VietEcon

I generally have a view that the efficient market hypothesis is correct, at least in its weak form. This basically means that most stock prices are random and that it is very difficult if not impossible to predict how they will move. But of course, the whole active management fund industry is built in opposition to this.

As a corollary to my belief, I also think that developed markets (the US, most of the European exchanges) are more efficient than stock markets in developing countries and these more efficient than frontier markets. Literature on the subject seems to support this view, but it is also based on my experience living in the Middle East for 10 years working as an equity analyst, although maybe I am just talking my book. As Upton Sinclair wrote: “It is difficult to get a man to understand something, when his salary depends upon his not understanding it!”

Anyway, I was interested to read a recent paper about the Ghana stock market. From the abstract:

The random walk hypothesis (RWH) was tested using four robust statistical tests, namely the Ljung-Box autocorrelation test, unit root tests, the runs test, and variance ratio tests (such as Wright’s rank and sign and Lo-Mac Kinlay). The empirical results showed that all four tests rejected the random walk hypothesis required by the weak-form efficient market hypothesis in all four return series. This provides empirical basis to infer that the GSE is inefficient at weak-form.

There have been multiple similar papers for other countries that show similar results, mostly in emerging markets, including for Vietnam, but mostly ending in 2013. I thought it would be interesting to try the same for the Vietnam stock exchange using the past five years of trading and see if it is “inefficient”. I didn’t have the time to run all of the statistical tests, so I did a quick and dirty regression of a few things to see if it was worth it to follow up.

An efficient market is one in which “successive price changes in individual securities must be independent.” I decided an easy test would be to look at successive prices (n, n+1) to see if there was any statistically significant explanation between the two. Basically, if a stock price is up on day 1, does that mean it would also rise on day 2. The way to test that is to run a Durbin-Watson test to see if “errors” are correlated. Basically, do a regression, find out what the fits the line and then find the errors for each of the points in the series. Then take those errors and see if one is correlated to the next. I did this for three different securities: the index (VNI), the largest cap stock (VIC) and a stock with a smaller capitalization but good volume compared to other small stocks (FLC).

P VALUES MOSTLY NOT SIGNIFICANT, R SQUARED IS SMALL AND DURBIN-WATSON TEST SHOWS NO AUTOCORRELATION

I was surprised to find that the regression showed that the successive price changes were independent. For both VN and VIC, the R squareds were minimal and the p-values were way out of significance range. The Durbin-Watson test showed the same - the results were clustered around 2, which means there is little autocorrelation.

I was somewhat surprised by the p-value for FLC, which was much lower than the others, meaning that the finding is significant at the 10% probability range. I ran these as the natural log of the change in price (ln(p+1/p). If I had run it as just as the percentage change ((p+1/p)-1) the p-value would have been just under 0.05, or significant at the 5% level.

The conclusion I take from the more significant p-value for FLC is that maybe smaller cap stocks are less efficient. Investment managers might be able to drive investment decisions on that basis, if the volume is there. The problem, though, is that for many even medium-sized funds, a stock that has a market cap of just USD161m may be difficult to invest in. Say you had a fund of USD200m to invest in Vietnam and wanted to take 20 positions with an average of USD10m, then you would own 6.2% of FLC. Even with solid volume, it would probably be hard to get in and out in a quick fashion.

Back to the market efficiency, I think it would make sense to run more of these tests - there are four that people usually run. We might find some cases where the market isn’t efficient that could help result in investment decision.