“…In fact, studies that use the growth rate in output or in per capita output during the election year as the main economic determinant of the incumbent government's electoral success, either finding or assuming growth in earlier years to be irrelevant, abound in the literature. Beside Fair (1978Fair ( , 1982Fair ( , 1988Fair ( , 1999Fair ( and 2004, these include time-series studies by Lewis-Beck and Rice (1984a), Burdekin (1988), Gleisner (1992), Chappell and Suzuki (1993), Rosenthal (1993, 1996), Alesina and Rosenthal (1995) and LewisBeck and Tien (1996) on U.S. presidential elections, by Kramer (1971), Lewis-Beck and Rice (1984b), Kiewiet and Udell (1998), and Grier and McGarrity (2002) on U.S. congressional elections, by Lewis-Beck (1997) on French presidential elections, Akarca and Tansel (2006) on Turkish parliamentary and local administration elections, cross-state time-series study by Peltzman (1987) on U.S. gubernatorial elections, cross-state study by Blackley and Shepard (1994) on a U.S. presidential election, election, pooled cross-national time-series studies by Powell and Whitten (1993) Peltzman (1990) analyzing U.S. presidential, senatorial and gubernatorial election outcomes, using pooled crossstate time-series data, and Abrams and Butkiewicz (1995) analyzing the outcome of a U.S. presidential election, using cross-state data, concluded that voters consider information from the incumbent's whole term, not just its final year. Their results nevertheless indicate that voters give relatively more weight to recent past of an administration than its distant past.…”