I thank Ellis Krauss for his thoughtful response. I try to answer each of his
questions in turn, placing my answer after his questions below.

On Mon, 18 Sep 1995 "ELLIS S. KRAUSS" wrote:

>1)I think I see what you are getting at by trying to find predictibility on an
independent variable, but I still have the question, if you are not trying to
explain the dependent variable [policy or policymaking] then what is the point?
You can't explain an independent variable [that's a contradiction!]. So what are
you trying to explain?

The word "explain" is the problem here. We still set up models to "explain" the
dependent variable. My point is that we often are less interested in a
fully-specified explantion (accounting for 100 percent of the variance in the
DV) than we are in showing that one or more independent variables are important
contributors to that variance. In statistical language, we're looking for high
t-statistics, not high R-squared.

I'll give a different example. Let's suppose that budgets are decided by taking
last year's budget and choosing to either reduce or increase it, item by item.
If this is so, then almost all of this year's budget is accounted for by last
year's budget (you'll get a huge r-squared by simply regressing budgets against
their lags). But we may be interested in the political decision to raise or
reduce each particular item -- the marginal variance. Say we hypothesize that
partisan change in government from Dems to Reps will cause defense items to
increase and welfare items to decline. If the t-stats on the partisanship
indicators are high enough, we believe that they help to explain the marginal
variance in budget sizes. Clearly, partisanship alone doesn't "explain" the
actual budgetary level for any program -- most of that is "explained" still by
last year's budget [and other things like inflation and unemployment]. But we
were interested in the effects of partisan change on budgets, not in why each
budget item is at its current level, so we're happy with the results.

>2)One reason I think you got so much flak on your previous postings is because
everyone takes it for granted that pols have formal authority to make formal
decisions in policy. The question, and I gather we all agree at least on this
part, is how much "slack" there is between that formal authority and its
enforcement/monitoring, etc. by the pols. Whether one phrases it as discovering
the extent to which pols actually exercise their formal authority given the
formal rules, or one phrases it as how much influence do pols actually have, I
think still we are not talking about different things.

Very well put. I think you're right about that. However, my point was that
framing it the way you just did -- how much slack? -- tells us what we should be
looking for in terms of causes of slack (e.g., how much administrative structure
and process to be used for monitoring, and how easy it is for pols to undo what
burs did, ex post), and how to test whether bureaucrats who "do all the work"
are really able to do whatever they want, or, alternatively are simply
productive and compliant agents. No longer will looking to see who does the work
suffice as an explanation of who is making the important decisions. We actually
have to start figuring out what pols want out of policy, whether they get it,
and why they do or don't.

But if
>formal authority is a constant, and there is empirical evidence of a change in
the influence pols are having on policy, then I think most of us are saying
something must have changed to give the pols more influence. I gather you say
that this can't be true if the rules haven't changed? So you either say, "show
me where the rules changed" in the sense of how formal authority has changed, or
deny that any change could possibly have taken place. Is that correct?

Right -- mostly. I don't believe (yet) that there is evidence that pols became
more influential in policy making after the oil shock. They always had that
influence. I do believe there is evidence that pols, specifically LDPers, became
more active in a public way in using their influence. They took positions, met
bureaucrats more often, formed study groups, discussed policy, informed
themselves, etc. And all of this is interesting and requires an explanation. But
that doesn't mean they ended up with a greater ability to affect policy than
before. It only means that they cared more or cared to make their input more
public than before. Perhaps this is because they had to fight harder to obtain
consistent portions of a shrinking pie --that makes some sense. But whatever
happened, they did not gain more power/authority/influence. I think you're right
that some of this is a semantic battle, and that some who observed the changes
after the oil shock meant no more than what I just said. "Power" is a slippery
word. That's why I try to be so explicit in defining what I mean when I say
authority, and why I rely on something easily measurable like the constitutions
and other laws that define the relationship between the Diet and the
bureaucracy. Naturally, formalism is only a starting point, and cannot explain
policy outcomes. However, it can, if you believe the formal rules of the game
are enforced -- another empirical problem -- tell us researchers where to start
in terms of asking who wanted what and why that might have changed. This leads
me to my answer to your third question, below.

>3)the analogy of the formal veto power of American presidents was interesting.
But if Carter exercised the veto more than Ford, then you are correct that the
formal veto power itself had not changed so it can't be the explanation. We have
to look for something that had changed other than the formal authority to
explain the variable. Doesn't this contradict your whole argument though about
pols in Japan? If the two are equivalent then looking for why pols have more
influence at time 2 than time 1 , despite having the same formal authority, is
the same as trying to find out why one President exercised veto power more than
another. No?

No. My point with the analogy is that we know that both Carter and Ford had
equal formal authority. One used one type of authority -- the veto -- more than
the other. The question is why. The answer I suggested was that Ford vetoed more
because he was given more than he disagreed with. This was because his Congress
was controlled by the Opposition Dems. Carter, who was a Dem also, disagreed
with less of the Dem Congress's output than did Ford, and hence used the veto
less often.

The point here is that the explanation for decline in the number of vetos from
Ford to Carter lies in the partisanship of the president (what he wanted
relative to what the congress wanted) and not in how powerful Carter was
relative to Ford. This is consistent with what I'm trying to say about the LDP
after the oil shock. I argue that the formal authority of pols did NOT change,
so the explanation for changed behavior must lie in preference changes, or in
changes in other constraints or incentives of pols which might induce preference
changes over behavior. So the bone I pick is with anyone who claim that pols
became more powerful in 1974. I say no, they changed their behavior for some
reason, but not because they suddenly had authority where they'd lacked it
before. Now let's figure out why they changed their behavior. Perhaps it has
something to do with the new constraints of a slower economy -- so that getting
bigger and bigger portions of pork for the home districts was no longer easy. I
don't know, but I know it isn't because of a power shift.

Thanks for the clarifying questions. I often bemoan the fact that we all are
forced to use words like "power," "rational," and "influence," etc. that already
exist in the common parlance but that mean different things to different people.
Most of what I've been saying in this forum is that we ratchers, for example,
don't use rational the way it's used in everyday language, to mean "correct," or
"reasonable" or "intelligent." Unfortunately, few of us are bold enough to
invent brand new words (e.g., Dahl's "Polyarchy") in order to eliminate the
confusion of words already in the lexicon. The best we can do is to try to be
precise about what we do mean when we use loaded terms.

--Mike Thies

---------------------------------------------------------------- Michael F.
Thies Department of Political Science
(ph) 310-825-1976 UCLA -- Box 951472
(fax) 310-825-0778 405 Hilgard Avenue
thies[atx]nicco.sscnet.ucla.edu Los Angeles, CA 90095-1472
----------------------------------------------------------------