Variable returns to scale (VRS) estimates are presented on the left side of Table 2. In
the non-restricted OLS specification we can observe that the sum of the coefficients is slightly
lower than one, supporting the assumption of decreasing returns to scale production
technology. To determine whether this outcome is statistically significant, we perform the
Wald coefficient restriction test. The associated F-statistics of the Wald test show that the null
hypothesis of CRS can not be rejected even at the 5 percent level. In addition, the ARMA
(1,1) specifications illustrated in columns (3) and (5) yield almost the same output elasticities,
which sum to unity, whether or not the CRS restriction is imposed. These results give strong
evidence on the existence of CRS production technology in Chinese industry over the period
1952-2005.