Nothing is wrong with your code. If you run the simulation several times, you'll see two significant spikes in acf; one positive and one negative. Make your sample size larger, say 1000, it may help.
Most of the time series in economics/finance are nonstationary, therefore, arima can be use to determine the number of unit roots in the process. arima (p, 1, q), arima (p, 2, q), etc.