Monday, 23 December 2013

Academic Update AY2013/14 Sem 01

The workload for this semester is relatively light. SAP is within my expectation, but was pleasantly surprised by my A in ST3246 as the final was really horrendous. This semester will also be the first time I'm going to use my S/U. I still can't believe I need to S/U a statistics GEM haha... maybe qualitative mod is really not my thing. After S/U, my SAP for this semester will be 4.875. Hopefully, I would be able to get a Dean's List this time.

ST1131 Introduction to Statistics A-
MA1101R Linear Algebra I A
MA1102R Calculus A
EC1301Principles of Economics A
GEK1508Einstein's Universe and Quantum WeirdnessA
CS1010 Programming Methodology A-
ST2131 Probability B
MA2108 Mathematical Analysis I B+
EC2102Macroeconomic Analysis I A
PC1322Understanding the UniverseA+

ST2132Mathematical StatisticsA+
ST2137 Computer Aided Data Analysis A+
EC2101Microeconomics Analysis I A
LSM1302Gene and SocietyA+
GEK2503Remote Sensing for Earth ObservationA-
ST3131Regression AnalysisA
ST3236 Stochastic Processes IB+
ST3239 Survey Methodology A-
EC3312Game theory & its application in economicsA
EC3333Financial Economics IA+

ST3233Applied Time Series Analysis A
ST3246 Statistical Models for Actuarial ScienceA
EC3361Labor Economics IA+
EC3383Environmental EconomicsA-
GEM2901Reporting Statistics in MediaS/U


The module is taught by Dr. Alkema Leontine. She is really passionate in her teaching and very approachable for consultation. Module is not difficult, but it may take some time getting used to the various types of time series. Workload is light with one take-home mid term and a project. The only complaint I have is that she released the details for the project very late into the semester despite covering all the necessary content a few weeks prior to that. Finals was very do-able, the only challenge is probably to compete all the questions within the time limit as I overheard a few person saying that they did not have time to complete it.

For some reason this module is extremely popular. I spent over 1000 bid points for the module as I'm really determined to get it because it will help me prepare for my SOA exam in February. I'm not sure why others are so keen in it tho. Perhaps, because it is only offered once every 2 years, or because of the word 'Actuarial' in it (Taking this mod, however will NOT help you break into the actuarial career if you are not intending to take the professional actuary exams).

The module is taught by A/P Lim Tiong Wee. A good lecturer and extremely patient when going through his tutorials. However the pace for tutorial can be quite slow sometimes. The workload is extremely light for this module with an average of 3/4 questions per tutorial and a total of 6 assignment questions which are a give away. The finals is, however, a completely different story. While his assignments focuses on calculations and application, his final is purely theoretical. I believe many, including me, are caught off guard by it. It also does not help that the paper is extremely difficult and I can barely do half of it.

This module is more qualitative than quantitative. Dr Chua Tin Chiu who is also our Vice Dean, spent most of the time sharing with us his experience as a statistics/survey consultant. Although his stories are very interesting, it gets tiring after awhile. Workload is light, no tutorial, no mid term, only 2 very simple 800-1200 words essay. 

Taught by Peter James McGee. Most of the concepts in the first half of the semester is similar to EC2101. The mid term is straight forward while finals have some challenging questions. Also, McGee like to set funny questions which often involve some extra-terrestrial beings. 

Mid-Term median: 72/100

One of the worst module I have taken. Lots and lots of readings. I was caught off guard during the mid terms as I was not expecting it to test so much on readings, and got a less than desirable score. Having learnt my lesson, I spent considerable amount of time and effort going through all the boring  readings to prepare for my finals, BUT NOT A SINGLE QUESTIONS ON READINGS CAME OUT. Although Alberto SALVO is an Engineer during his undergrad years, his module is highly qualitative. Will not recommend this module to those who prefer modules which are more quantitative.   

Mid-Term median: 78/100

As usual, if anyone would like to have materials such as lecture notes or tutorials for any of the modules that I have taken before, just leave a comment below or email me ( with your nus email. Also, a simple thank you would be appreciated when I send you the materials.

Wednesday, 4 December 2013

Mutual Funds: An empirical study

The purpose of actively managed funds is to outperform the broad market index by exploiting informational gaps to identify any undervalued or overvalued stocks. A seemingly contradiction to the Efficient Market Hypothesis proposed by Eugene Fama in the 1960s who argued that it is impossible to beat the market because the stock market is efficient and all information are reflected by the stock price. Of course, theoretical hypothesis seldom model what we observe perfectly. In reality, market inefficiencies exist. Investors are influenced by many psychological factors which cause the prices to deviate from their intrinsic value. But how successful were the fund managers in exploiting these inefficiencies? To answer the question, several studies on the (relative) performance of actively managed funds were conducted. The following quotation summarizes their findings:

  • "Eugene Fama, William Sharpe and Jack Terynor were some of the first researchers to note the apparent lack of skills by mutual fund managers."
  • "Economist Michael Jenson provided his view in 1967… there is very little evidence that any individual fund was able to do significantly better than that which we expected from mere random chance ."
  • "Carhart (1997) observed that although some mutual funds outperform, on average, mutual fund managers did not exhibit superior investment skills.” 

This blog entry will highlight some of the findings in a more recent study conducted by Ferri and Benke, which investigated the probability that an index fund outperforming actively managed funds. Unlike previous studies, Ferri and Benke had taken precaution against survivorship-bias in actively managed fund which tend to positively skew their performance. Several restrictions were also introduced to the study which penalizes the performance of index funds, including the omission of front or back-end loading from actively managed funds, index funds with highest expense ratio was used when two or more share classes of the fund existed and pretax performance was used (index funds tend to have better tax efficiency). 

Findings # 1
Index funds have a higher probability of outperforming actively managed funds when combined together in a portfolio.
Based on the fund performance between 1997 and 2013, three-fund index fund portfolio has a probability of 82.9% in winning a comparable actively managed fund in term of returns. It is also worth noting that among those which outperformed the index fund, the median excess return is only 0.52%, compared to the median excess return of -1.25% of the underperformers. Probabilistically, it does not pay to invest in actively managed funds.
Figure 1 illustrates the result. The excess returns of actively managed funds were arranged from worse to best. Returns below 0 indicate that the index fund had performed better than the particular actively managed fund and vice versa. 

Findings # 2
The probability of index fund portfolio outperformance increased when the time period was extended from 5 years to 15 years
Given the difficulty to beat the index fund, perhaps it is not surprising that the longer the investment period, the harder it is for an active fund to continuously beat the market. Ferri and Benke study goes on to show that this hypothesis is supported empirical results. On average, the index fund outperforms the active fund 76.5% of the time in a 5 year period and 83.4% in a 15 year period.

Findings # 3
The probability of index fund portfolio outperformance increased when 2 or more actively managed funds held in each asset class
A surprising finding in the study is that diversification within asset class in actively managed funds actually hurt the investor. When each asset class is diversified into 2 or 3 active funds, the probability of index fund winning increases to 87.1% and 91% respectively, compared to 82.9% when there is no diversification.  

Findings # 4
Fees affect performance only to a small degree.
A major criticism of actively managed fund is the high fees and charges payable by the investors. It includes one off charges such as front-end load, back-end load which ranges from 1% to 5% and switching fees which are typically around 0.5% to 1%. In addition, there are also recurring fees, known as Total Expense Ratio (TER), which are payable annually, such as the management fees, trustee fees and miscellaneous fees which range from 0.5% to 2%. All these charges/fees can reduce the returns of your investment substantially. However, based on the empirical results of Ferri and Benke study, the effect of fees on fund performance is not significant. Although, it should be noted that one-off charges are omitted in the study.

The findings presented overwhelming evidence against actively managed funds. While the findings certainly do not preclude the possibility that any particular actively managed fund will outperform the index fund at a given point in time, predicting the ‘correct’ fund is highly unlikely and even less so with a long time horizon. Investors, especially the less sophisticated, might have a higher chance of realizing their investment goals using the simple index fund strategy. 

For those who are keen, the original report is available here: A Case for Index Fund Portfolios.