Appendix A
A Simulation Analysis of Potential
Future Borrowing
A set of simulation projections helps to illustrate the risk of insolvency
posed by low trust fund balances at the start of 1997. All projections are based
on the state-level relationship between increases in unemployment and de-
creases in the reserve ratio multiple during 1990-1992. A cross-section regres-
sion with weighted state data was fitted to derive parameters for the
relationship. The resulting equation was then used to project counterfactual
trust fund drawdowns during 1990-1992 and 1997-1999. For both periods,
initial reserve ratio multiples were noted and alternative recession-related
drawdowns were simulated. The results show that many more states could
need loans than the seven that actually borrowed during 1991-1995.
The cross-section regression, with states weighted by the size of their labor
forces, was the following:
Eq. Al RRM92 - RRM89 = 0.3157 - 0.5526 x (TUR9092/TUR8789)
(2.65) (6.01 )
Adj. R2 =0.8) 8
Std. err. = 0.249
where RRM89 and RRM92 are the reserve ratio multiples, TUR9092 and
TUR8789 are average unemployment rates for 1990-1992 and 1987-1989,
and t ratios appear in parentheses beneath the coefficients.
The regression shows a strong statistical association between the two se-
ries, with larger decreases in reserve ratio multiples for states in which unem-
ployment rate ratios were higher.
Chapter 1 notes that seven states needed U.S. Treasury loans during the
1991-1995 period. As shown in Table 1-4, Connecticut, Massachusetts, Mich-
igan, and the District of Columbia had negative net balances at the end of 1992.
Maine, New York, and Missouri had positive net balances at the end of 1992,
even though they needed loans. Note, however, that the two states with large-
scale borrowing during 1991-1994 (Connecticut and Massachusetts) had neg-
ative net reserves at the end of 1992, and their reserve ratio multiples were the
most negative of all state-level multiples.
A simulation that projects reserve ratio multiples at the end of hypothetical
recessions would not be expected to identify all states needing loans. Cash-
flow loans would be expected for some states that ended the year with positive
trust fund balances. Further, the simulations do not attempt to measure the size

-149-