Moving average increases autocorrelation?

[musicgold]

Hi,

I have a data series of monthly change in the 1-year government yield. The series some auto-correlation- lag 1 and lag 2 had correlations of 0.20 and 0.23 respectively. After that the ACF dies down quickly and oscillates around the zero level.

To eliminate that auto-correlation, I converted the series into a 3-month moving average, but that has actually increased the auto-correlation of the time series. I am not sure why the auto-correlation is increasing.

Thanks.

 

[passyingby]

t = your month.
Y = difference in yield between t and t-1

The correlation between (Y[t] + Y[t-1] + Y[t-2])/3 and (Y[t-1] + Y[t-2] + Y[t-3])/3 is going to be pretty high because they 2/3 of the data points are the same!