Rational Prospective Citizens Partisan Cycles
Thus, unless winners were perfectly foreseen, actual election outcomes produce surprise policy, yielding short-term real-outcome shifts that fade as elections recede and contracts slowly adjust to the actual winner.7 Alesina and Rosenthal (1995) and Alesina, Roubni, and Cohen (1997) collect and advance Alesina and colleagues’ (see n. 2 above) political-economic cycles work, contrasting evidence in support of electoral cycles and partisan cycles, in their RE and non-RE variants, from political-economic data across postwar US and OECD democracy samples. They find the data consistently favor RE models, indicate strong partisan, although few discernible electoral, cycles in macroeconomic outcomes, and suggest electoral and partisan macroeconomic-policy effects. The wider empirical literature, however, is more equivocal. Recall that RE competence-signaling and non-RE electoral cycle theories have similar predictions, except that RE limit cycle size, regularity, or duration. The differences between RE and non-RE partisan theories of macroeconomic outcomes are also subtle: in non-RE partisan theory, policy-makers exploit stable Phillips curves to shift macroeconomic outcomes permanently. In RPT, only unexpected policy creates real effects, so when the left wins, inflation rises but real outcomes rise only insofar as the election winner and so the policy shift was unforeseen.
Post-election, expectations adjust to incorporate the higher inflation, returning real outcomes to natural rates over time as new contracts come online, but inflation remains higher. The opposite pattern unfolds if the right wins. Thus, RE and non-RE partisan theories differ primarily in whether real partisan effects persist or fade. In US and/or OECD data, Alesina and colleagues (see n. 2 above) find indicators equal to (1,−1) in the first 8± quarters (2± years) of (left, right) governments empirically dominate traditional indicators equal to (1, −1) over governments’ whole terms. They interpret this as supporting the short-term real effects predicted by RE models. By contrast, inflation is permanently higher under left governments in both RE and non-RE models—a pattern which the data, explored in this manner, also support. Empirical dominance of short-term indicators is indisputable; yet strongly concluding for RPT on this basis is premature (Franzese 2000).
For one, the estimated cycles differ little substantively. More critically, much besides RE could explain shorter-term partisan effects. For example, Alesina, Roubini, and Cohen (1997) describe how left parties first apply expansionary policy, then, observing rising inflation, a potential electoral liability, switch to contractionary policies; rightist governments proceed oppositely. These paths, which the mid-term balancing of Alesina and Rosenthal (1995) would also produce, for instance, yield shorter-term cycles; honeymoon effects would also, as would any diminishing returns from stimulation. Finally, and most troubling for RPT, substantively and statistically stronger US realgrowth partisan cycles emerge, before 1972, but significant right/left inflation differences, which produce RPT’s real cycles, emerged only after 1972.8 Furthermore, Alesina, Roubini, and Cohen (1997) also explore US money growth, nominal interest rates, budget deficits, and transfers, finding some, but weak, partisan differences in monetary policy, though evidence is stronger in 1949–82 in nominal interest rates. (Evidence of pre-electoral monetary policy effects is also lacking, but possibly only because exchange rate regimes were ignored in that analysis.) Problematically, the partisan indicator in these policy studies lags two quarters, but the real-effects findings mentioned above assumed lags of just 1–2, implying that real effects emerge before monetary policy changes. This sequence directly suggests Drazen’s (2001) activefiscal/ passive-monetary cycles in which fiscal policy drives partisan real-outcome cycles and nominal cycles arise from monetary accommodation of the fiscal manipulations. Also, the finding for monetary policy effects only in 1949–82 and the earlier finding of dampened inflation cycles through 1972 leave only a narrow 1973–82 window for partisan cycles to emerge from surprise inflation. Finally, the Phillips-curve slope required to produce the estimated real effects from the estimated monetary effects is implausibly steep.9 RPT also predicts more specifically that real-outcome effects should be proportional to electoral surprises.
Alesina, Roubini, and Cohen (1997) very cleverly apply option-pricing theory to measure electoral surprises and find partisan unemployment effects in monthly US data proportional to these measures. However, they test for cycles proportional to surprise size (so measured) only against the absence of cycles; i.e. the alternative is no partisan effect rather than a simple partisan cycle. Moreover, RPT cycle amplitude actually depends on electoral surprises times expected inflation differences between incumbents and challengers rather than just the size of electoral surprises, as these models specified. The empirical models thus implicitly assume equal incumbent–challenger polarization across all elections. This would bias estimates if, e.g., victory probabilities relate to distances between candidates, which they would in RPT or most other reasonable models. Data from prior-office voting records of most presidential candidates, which could gauge the requisite distances, exist.
Of course, the issues raised above—missing policy links behind the observed policy cycles, congressional influence, and varying exchange regime—apply to this analysis also. Summarizing, Alesina and colleagues (see n. 2 above) clearly establish shortterm real-outcome and long-term nominal-outcome partisan cycles, but the RPT model’s explanation for these cycles is not well established empirically. In particular, monetary policy and nominal-outcome patterns cannot easily explain real-outcome patterns. Regarding policies, Alesina and colleagues find strongest partisan effects in two-party/bloc countries—which is intuitive since these systems produce greater right–left government alternations—and in redistributive policies like transfers—also completely intuitive. They find only weak signs of pre-electoral tax manipulation and, weaker still, of pre-electoral spending manipulation..
