Suppose that daily changes for a portfolio have first-order correlation with correlation parameter 0.12. The ten-day VaR, calculated by multiplying the one-day VaR by the square root of 10, is $2 million. What is a better estimate of the VaR that takes account of autocorrelation?