Modelling Philippine Markets
By: Luigi Cortez
Associate Director, Risk Quantitative Finance & Economics, MCG.
Any firm with positions in financial instruments is exposed to market risk. This is because current accounting practices dictate that financial instruments be reported according to fair market value. This practice, called marking to market, exposes holders of financial instruments to market price movements. Market risk, the risk arising from these movements, must be managed effectively because financial markets are very volatile. Large movements are relatively common and strike unexpectedly.
For instance, on 19 October 1987, the US stock market plunged by more than 20%, wiping out more than $1 trillion worth of capital. The Japanese Nikkei stock index slid down from 39,000 in 1989 to 24,000 in 1 year, and further down to 17,000 in 3 years. Losses were estimated to be around $2.7 trillion. In the late 90's and early 2000's, the Philippines faced a currency crisis. From P26.3/$ at the end of 1995, the exchange rate climbed to 37.2 by the end of 1996, to 40.6 in 1999, 49.9 in 2000, and peaked at 56 in 2004.
Market risk is relevant not just for investors but also for the economy as a whole since some of the largest holders of financial securities are financial institutions themselves. Regulators have sought to address this by requiring banks to set aside capital commensurate to risk exposure. This serves two purposes. First, if extreme movements occur, losses can be shouldered without risk of insolvency. Second, such capital charges hopefully deter banks from excessive risk-taking. This naturally raises the question of how much capital should be set aside. Holding too little for a given risk will not provide much protection but holding too much will be detrimental as well.
Market risk is measured by what are known as risk measures (e.g. VaR and Expected Shortfall). Many more can be constructed and used in practice but what all of them have in common is that they are forward-looking. Because what happens in the future cannot be known today with certainty, risk measurement entails taking a stance on how assets prices will behave in the future. This question of how asset prices can reasonably be expected to behave is the primary motivation for modelling market risk factors.
We focus on modelling equities (stocks), foreign exchange (FX), and interest rates (for bonds) as these are the most significant sources of market risk in the Philippines.
Our modelling approach will be multidisciplinary. We place emphasis on qualitative economic assessments and use them as foundations for our quantitative models. The qualitative aspect takes precedence because it can stand on its own. For instance, while it would be helpful to have enough capital to withstand a currency crisis, it would be better to avoid losses in the first place.
Below are some qualitative assessments we want to capture in our models. First, financial data is inherently noisy, even under normal market conditions. Because of the large number of market participants, prices of financial assets fluctuate not only from day to day, but even from minute to minute. Any model of prices should take this noise into account.
Second, price volatility varies over time. Prices can fluctuate wildly in some periods but remain relatively stable in others. Additionally, high volatility periods tend to occur together, in a phenomenon called volatility clustering. More often than not, they represent crisis times and other idiosyncratic shocks to financial markets.
Third, we express some medium- to long-term views on three major asset classes in the Philippines. For the currency market, we expect the peso to continue appreciating relative to the U.S. dollar and do not anticipate large sudden movements. These are attributable to stable economic conditions and fundamentals. For the stock market, we believe equities are currently overvalued and anticipate downward corrections in the short to medium-term. Lastly, because current economic conditions call for tighter monetary policy, we anticipate upward movements in interest rates, translating to losses on bond holdings.
Because of the numerous factors in financial markets, price movements can be modelled using stochastic (random) models. This isn't to say that price movements are actually random; prices are deterministically set by pairs of buyers and sellers. However, in financial markets, there are just too many deals and factors to take into account. These lead to the noisy behavior of asset prices. Consequently, they can be modelled as if they were random.
We entertain a variety of quantitative tools to model asset prices. These include time series econometric models, regime-switching GARCH models, extreme value theory, state-space models, Kalman filtering, copulas, and Monte Carlo simulations.
We stress at this point that quantitative models in finance do not describe to us how reality actually works; they merely provide sketches. The most our tools can do is replicate certain statistical and behavioral properties in the data set chosen by the analyst so that even model and data choice express an implicit view on the markets. Modelling is subjective because it is tantamount to saying, \We expect the markets to behave like this." Therefore, results should not be taken at face value.
Qualitative assessments form the core of the risk modelling process and quantitative models translate it to information pertinent for decision-making purposes. Because of this important interplay, a multidimensional approach for risk modelling is indispensable.
Withdrawal Rates Study
By: Sami AlMuelen
Withdrawal rates reflect the rate at which employees leave the company, for reasons of resignation or retirement. These rates vary according to external factors, specifically, Industry patterns (e.g. BPOs), as well as to internal factors, given the Company’s overall strategy, cascaded down to its HR policies. The rates may thus vary according to years of service and/or age.
This Study was conducted to survey the withdrawal rates of our clients, with the objective of, to the extent possible, accurately reflecting our clients’ experience, according to industry. We are recommending the use of the appropriate set of rates, representing the “best estimate” for the underlying withdrawal assumption in our actuarial valuations.
The withdrawal or turnover assumption is a significant actuarial assumption affecting the cost of a retirement plan. If the company experience, with regards to turnover, is high, then the projected retirement costs will be adjusted to reflect this cost savings, to the extent that there is a probability for employees not remaining in employment to receive some form of benefit from the retirement plan.
Our recommended rates prescribe tailor-fit rates for each industry, given the adequate sample experience, and were computed using Kaplan-Meier type estimators for large data sets. They were tabulated by age, and by years of service, with separate tables made for each industry. Graduation of each these tables was done using the Whittaker- Henderson-Lowry method, with the parameters used for the said graduation fully disclosed in the paper.
Philippine Retirement Liabilities and Assets Survey for 2012 and 2013
By: John Alexander Arthur AASP
The Bangko Sentral ng Pilipinas (BSP) publishes total assets under investment management activities intended for employee benefits. For the years 2012 and 2013, total assets were indicated to be P 292.17 Billion and P 310.05 Billion, respectively. The published data were in terms of total assets under management as of year-end but did not, however, reflect neither cash flow nor gains or losses for the period. In assessing whether retirement funds are being managed effectively from a risk-based perspective, we performed a study to capture the movement in the assets and, consequently, in the total retirement liabilities themselves.
In order to achieve the above objective, a purposive sample1 of 438 audited financial statements for the year 2013, pertaining to retirement plans of top Philippine companies, was taken in order to estimate movement in retirement plan assets and liabilities. The distribution of assets was estimated as well.
Even though the sample is purposive, quasi inclusion probabilities proportional to size are assigned to invoke estimators for samples with unequal probabilities and bootstrap resampling procedures were used to measure the precision of these estimators.
The study revealed some pertinent results:
- Between 2012 and 2013, the unfunded retirement liabilities rose significantly in terms of percentage and absolute amounts. This may be primarily attributed to the decline of coupon rates provided by government securities.
- In addition to the decline of government securities’ rates, most companies however appear to be under contributing to their retirement plans, even though this implies missing out on opportunities like tax deduction and possible higher compounded investment gains.
- As compared to advanced and emerging nations, the asset distribution of retirement funds in the Philippines is too concentrated on fixed income and cash.2 Given that the Philippine population is relatively young, companies have a longer investment horizon and may absorb the associated short-term risk from investing in equities and alternative assets in exchange for maximizing returns.
Given these results, we recommend that companies engage in an actuarial funding valuation in managing their retirement plan costs, rather than solely relying on the IAS19 valuation for accounting disclosure purposes.
The IAS19 valuation presents the retirement costs as they accrue. With a young population, these costs are initially low but inevitably will spike upwards as the population ages. And the costs may spike at a time when cash flow could be tight. The funding valuation, on the other hand, provides a long term perspective of the costs and aims at spreading these costs evenly into the future. Because of this level funding approach, initial contributions may be generally higher than under the IAS19 valuation but this eventually translates to higher compounded investment gains in the future, thus reducing the Company’s overall cash layout.
For a hypothetical retirement plan whose population average age is 30, the fund composition at the end of a 30-year investment horizon under the two funding methodologies has been simulated, the results of which appear below:
||Retirement Fund Composition
|Funding (EAN) Method (Level Funding Approach):
|IAS19 Valuation (Cost funded as accrued):
Note that under the level funding approach, only 31.76% of the total retirement obligation is contributed by the employer. The remaining 68.24% is derived from the investment earnings.
In conclusion, we wish to point out two important advantages:
- A contribution strategy based on the funding (EAN) valuation method requires less in total employer contributions, compared to that under the IAS19 contribution strategy, in funding the total retirement obligations.
- An actuarial valuation, i.e., IAS19, should not be undertaken for merely disclosure compliance purposes but, based on whichever funding valuation method is preferred, a contribution strategy should actually be implemented by making the retirement contributions, in the amount and timing recommended in the actuarial valuation report, in order for the employer to reduce costs because a portion of the total retirement outlay will be coming from the investment earnings.
Philippine Retirement Liabilities and Assets Survey for 2013 and 2014
By: John Alexander Arthur AASP
Between 2013 and 2014, the assets managed by Universal and Commercial Banks intended for employee benefits rose from P 310.05 Billion to P 350.03 Billion. This translates to a 12.89% increase in assets being managed for employee benefits. This is a significant increase considering that the increase between 2012 and 2013 was only 6.12%. This is testament that the revision of the IAS 19 in 2013 had significantly impacted how corporate leaders give regards to topics about private retirement, whether on a financial or administrative aspect.
However, the continuous decline of government coupon rates on later tenors increases the present value of retirement obligations. There is a case for revisiting the index being used for discounting retirement benefit payments. One may argue that the plan assets should have increased due to the decline in the discount rate, but the portfolio mix of the retirement plan assets had changed over the course of a year.
A purposive sample of 288 audited financial statements for the year 2014 pertaining to retirement plans of top Philippine companies was taken in order to estimate movement in retirement plan assets and liabilities. The distribution of assets are estimated as well. Quasi inclusion probabilities proportional to size are assigned to invoke estimators for samples with unequal probabilities and bootstrap resampling procedures in measuring precision.