17
addition, rival fund expenditures has a statistically significant negative impact upon LDP
vote shares, or a loss of 3% of the vote share for each additional ¥100 ($.94) spent by the
main rival (0.03 for ¥1).
16
These models suggest the importance of “own spending” effects for the LDP upon
vote shares. All of the “own spending” variables—both the campaign period and year-
based ones—are statistically significant and positively correlated with LDP vote shares.
These results support Hypothesis 1, which postulates that own spending should positively
increase vote share. However, Hypothesis 2, which states that the effects of rival
expenditures should be negatively correlated, is not supported for local party branch or
koenkai expenditures, but only through the results of fund agent spending reported in
Models II and III.
As rival fund agent spending appears to have a small effect on LDP vote shares, I
examine whether there are any significant simultaneity effects between own and rival
expenditures and vote shares. Using path analysis and standard OLS regression, I did not
uncover any statistically significant indirect effects between rival expenditures, own
expenditures and vote shares for the LDP.
17
How much LDP candidates spend during the
year-based period does not appear to be related to the expenditures of their main rival.
However, the year-based expenditures of their main rivals do have a small statistically
significant effect on their district share of the vote as demonstrated in the above models.
16
The coefficient for rival campaign-period expenditures also show a statistically significant positive effect
on LDP vote shares, although there is no theoretical expectation for this result.
17
I examined the effects of total rival spending on the total expenditures of LDP and found no statistically
significant results for either party (results available from author). In a standard OLS regression of rival
spending on own spending, the coefficient was insignificant at 0.00(.10) for the LDP and 0.04(0.03) for the
DPJ, mirroring similar inconclusive results using path analysis.
Convention