i have a data of stock prices in daily frequency. i want to study the relationship of stock price(or returns) with select macro-economic variables. the macroeconomics variables are in monthly series.
so, i have to make the daily frequency of stock prices as monthly frequency.
if i calculate average, i doubt whether it will be representative or not, becuase of the longer time period(ie., one month) and during the month, there may be some extreme values in the distribution
what the the appropriate method in this regard?
thanks in advance
Alternatively, you may consider an econometric approach so-called MIDAS (mixed-data sampling), in order to combine the variables in different frequencies in your model. Ghysels developed this technique and made huge theoretical contribution in this area. As an example, if your regressor variable appears at a higher frequency than the dependent variable, then you can combine these variables in different frequencies in your model without finding the representative value.
You may have a look at the following articles:
Ghysels E, Santa-Clara P, Valkanov R (2006). Predicting volatility: Getting the most out of return data sampled at different frequencies. Journal of Econometrics, 131, 59–95.
Ghysels E, Sinko A, Valkanov R (2006). MIDAS Regressions: Further results and new directions. Econometric Reviews, 26, 53–90.
Andreou E, Ghysels E, Kourtellos A (2010). Regression Models With Mixed Sampling Frequencies. Journal of Econometrics, 158, 246–261.
Ghysels et al. (2016). Mixed frequency data sampling regression models: The R package midasr. Journal of Statistical Software.
https://www.jstatsoft.org/article/view/v072i04
Dear Srikanth
I would calculate the stock returns using end of month prices plus any dividends paid during the month. The series of monthly macroeconomic variables should be converted, if they are not already in this form, to monthly changes in the dependent variable-series. Then you have all the variables in the same format (monthly changes).
Hi, srikanth,
There is a very simple way that is to program (on XL) to select a day in the middle of the each month (to avoid the anomaly of the end-of-month effect) and to take yield of this date.
Raymond A K Cox's comment is right. The standard procedure is to take the last data of each month. It is not correct to use the average of the daily prices of each month as a month's representative value. Regards,
we generally take month end day closing price, but this may affected by month end effect anomaly. and more over we consider that firm didn't pay any dividend during the month.
Dear Hamid
There is a turn-of-the-month effect on stock returns, e.g. Article Seasonal Movements in Security Prices II: Weekend, Holiday, ...
However, the bias can be ignored if the study is for a long-term and consistently uses end -of-the-month stock prices.
Dear Mr. Potharia,
Since you are doing time series analysis another suggestion would be to use ARIMA modelling. You can refer to our paper Mallick et. al. (2007) on Mathematical Statistical Pricing...... on www.researchgate.net/profile/Soumitra_Mallick.
S.K. Mallick
for RHMHM School of Mathematical Sxciences Thinking USA, Japan, India (proposed)
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India