The paper investigates individual earnings and hours using Japanese panel data. We find that (1) the variance of income changes of Japanese households is approximately one-third of that of the US; (2) the autocovariance of income changes at first lag is approximately one half of the income variance; (3) the autocovariances of income changes at lags more that one year are negligible; and (4) the cross-covariances of changes in income and hours worked are very small. The decomposition of the variances into permanent shocks and measurement errors reveals that one-third of variations of income changes are due to permanent shocks.