Explore - Experience - Excel

Semi-Efficient Market Hypothesis – Does It Hold Good?


“A blindfolded monkey throwing darts at a newspaper’s financial pages could select a portfolio that would do just as well as one carefully selected by the experts” – quotes Burton Malkeil[i]on market efficiency theory. Efficient Market Hypothesis, term coined by Eugene Fama[ii]way back in 1970, and further studies in the area went onto earn him the Nobel. The theory stated that markets are efficient or intelligent enough to subsume all the information relating to firms in general and the price discovered by market is the right price. When, market has used every bit of information in arriving at the price, no one individual market participant can make an excess or abnormal return by beating the market’s intelligence. Only arrival of new information can lead to a change in the price, and the invisible angel in the market will very soon bring the price to incorporate the new information also and thus leaving no further speculative opportunity.

The theory is further enhanced by dividing the level of efficiency on the basis of the kind of information that could have been subsumed by the market. If we presume that a given market is using all the historical information relating to a scrip, we call the market to be efficient in its weak-form; if it is presumed that along with historical data, the market is also using publicly available information, including market sentiments, rumors and expert opinions, we call the market to be semi-strong efficient; and finally if we presume market has access to information that is privately known to only the insiders, and the price discovery mechanism uses such data also, we call the market to be efficient in its strong form. 
EMH being a theory, even though seems logical enough, it needs extensive research across markets, across time lines, and through different statistical tests. Hence, EMH studies are a regular research problems, being attempted by capital market researchers since decades and it continues till date.

Semi-Efficient Market Hypothesis 
efficient, or it is the individual stocks that are efficiently priced. It was then observed that most significant contributions to these theories have concluded based on their tests conducted on a market as a whole, and not individual scripts. Prasanna Chandra[i] puts it as below:

The efficient market hypothesis (EMH) has a cousin, the semi-efficient market hypothesis (SEMH). SEMH holds that some stocks are priced more efficiently than others. Consider two companies, Infosys and a hypothetical start-up firm called Nuvo Software. Infosys is followed by many investors and actively traded. Thousands of portfolios would contain Infosys shares and innumerable security analysts follow it. Hence it likely to be fairly priced. What about Nuvo Software? Very few people follow it; so according to SEMH there is a greater likelihood that it will be mispriced. Extending this idea, one may argue that the market perhaps has several tiers. Put differently, there is a pecking order of efficiency. Most observers of the market are generally sympathetic of the logic of SEMH”

Two observations from the above explanation by the author. One, it may make more sense to study individual stocks to determine the efficiency of a market. Two, efficiency of a market follows a pecking order and hence, it would be impossible for one to put a dichotomizing line between an efficiently priced and a not-so-efficiently priced scrip, as the pecking order is more discrete than continuous.

Sample study for SEMH 

We have conducted a sample study on 12 stocks from Indian markets, 4 each from large cap, midcap and small cap to analyse if the prices follow a pattern among these stocks. In essence, we are checking for the weak-form of efficiency (i.e., whether the market determined prices of these stocks are randomly priced, and if so, we conclude that market has used all the historical information, in the form of historical prices, and efficient in its weak-form) of each scrip. Based on the results we can derive inputs for SEMH theory. 

Stocks Selected 

Stocks selected using convenience sampling as below:

Large-Cap (forming first 75% of market cap): Infosys, BPCL, SBIN & Tata Steel (random 4 from CNX NIFTY) 

– Mid-Cap (forming market cap between 75th percentile to 90th percentile) : Petronet LNG, Reliance Communications, Marico Ltd, Pidilite Industries Limited (random 4 from CNX Midcap

– Small-cap (forming market cap between 90th percentile to 95th percentile): Cox & Kings Ltd., Gati Ltd., Raymond Ltd., Zydus Wellness Ltd., (random 4 from CNX Smallcap)

Methodology used 

As the objective is to test the random movement of the stock prices, a non-parametric runs test is conducted. Runs test follows the mechanism of designating a plus (+) sign when there is increase in price and a minus (-) sign when a decrease in price occurs, when the stock prices are arranged chronologically. A run occurs when there is no difference between the sign of two changes. When the sign of change differs, the run ends and a new run begins. The total number of runs of a series of stock prices are then tested for statistical significance (i.e., whether the runs of the given sample series statistically differs from the number of runs of a purely random series of the same size). Thus for our study we tested for randomness using runs test of the 12 selected stocks based on their weekly closing prices. The hypothesis tested hence for each script was in the below form:

H0: The weekly closing price series of Infosys is randomly distributed

H1: The weekly closing price series of Infosys is not randomly distributed

Test Conducted: One-Sample Runs Test, around the mean

                      Decision Criterion: Reject H0 if the sig. value is less than 0.05 (95%  con. level)


Test for randomness as above is conducted using SPSS 23.0 v. and the results are provided in the table below: 


– Null Hypothesis is rejected in ALL CASES, submitting proof to infer that “all the selected scrips, across market capitalizations, do not follow random walk, hence, none are efficiently priced in their weak-form”.
– Based on the above, we can also interpret that, SEMH also does not hold gold. This is evident in the sample case that we have tested above, where all the stocks have emerged to be inefficiently priced.


Contrary to the arguments of few researchers, on the pecking order of efficiency, above results deduce the absence of such order and also leaves ample scope for further stimulations to question the existence of SEMH.

It may be noted that the above conclusions are based on sample study. As has already been mentioned, either EMH or SEMH is still a ‘Hypothesis’ and demands further research using different sample, sample size, timeframe, testing tools and markets.

lkeil, 1973)
[ii](Fama, 1970)
[iii](Chandra, 2012)