Error in !class_ind : invalid argument type (pgmm问题)

问题1.## Blundell and Bond (1998) table 4 (cf DPD for OX p.12 col.4)
z2 <- pgmm(log(emp) ~ lag(log(emp), 1)+ lag(log(wage), 0:1) +
           lag(log(capital), 0:1) | lag(log(emp), 2:99) +
           lag(log(wage), 2:99) + lag(log(capital), 2:99),        
           data = EmplUK, effect = "twoways", model = "onestep",
           transformation = "ld")
summary(z2, robust = TRUE)

当我运行上面代码的时候，发现最后出现的结果是：Error in !class_ind : invalid argument type。没有估计出我要的结果，请问是什么原因呢？希望指点一下。

问题2.z2 <- pgmm(log(ISR) ~ lag(log(ISR), 1)+ log(ITD) +log(ED)| lag(log(ISR), 2:99),data = EDATA, effect = "twoways", model = "twostep",transformation = "ld")
summary(z2, robust = TRUE)

当我运行问题2代码的时候，发现结果错误提示：Error in solve.default(crossprod(WX, t(crossprod(WX, A2)))) :
  system is computationally singular: reciprocal condition number = 2.49463e-19
In addition: Warning messages:
1: In pgmm(log(ISR) ~ lag(log(ISR), 1) + log(ITD) + log(ED) | lag(log(ISR),  :
  the first-step matrix is singular, a general inverse is used
2: In pgmm(log(ISR) ~ lag(log(ISR), 1) + log(ITD) + log(ED) | lag(log(ISR),  :
  the second-step matrix is singular, a general inverse is used
原因应该是我数据本身的原因，希望有知道的能一起讨论下原因所在，不胜感激。

以上知道一个问题的都希望能给予回复下，感激不尽。