I came across a paper that write the following: The first set of independent variables, REPRESSION(1)i,t-1, REPRESSION(2)i,t-1, REPRESSION(3)i,t-1, REPRESSION(4)i,t-1, are binary indicators measuring a state's previous level of repression. They are included in place of the standard lagged dependent variable to account for dependence across the categories of the dependent variable over time. Because REPRESSIONi,t are treated as a nonlinear dependent variable, it is not appropriate to control for autocorrelation through the standard treatment of REPRESSIONi,t-1 as a lagged independent variable. The inclusion of four dummy variables is a nice alternative to the problem of correlated categories of repression within a state across time.

The model that the author uses is

REPRESSIONi,t = a + REPRESSION(1)i,t-1 + REPRESSION(2)i,t-1 + REPRESSION(3)i,t-1 + REPRESSION(4)it_1 + other control variables + error term.

The dependent variable REPRESSIONi,t is an ordinal variable, ranging across five levels of repressive behaviors.

Can someone explain what this means?.. Why does the author include 4 different dummy variables as an alternative?