I believe Thies is referring to TJ Pempel's article "The Unbundling of Japan,
Inc: The Changing Dynamics of Japanese Policy Formation," which was published in
an important book of essays from the folks at the Journal of Japanese Studies
called _The Trade Crisis: How Will Japan Respond_, ed. by Kenneth Pyle, 1987.
There are some other essays in somewhat the same vein by Haru Fukui and Michio
Muramatsu.
As the title indicates, this article argues that back in the 1960s, Japanese
policymaking was relatively well coordinated and consensual within the
"conservative camp" of the LDP, the economic ministries and peak business
interest groups, although there were sharp clashes with the opposition. In the
1970s things got much more complicated and conflictual within the conservative
camp, and the oppostiion was also more influential; there were attempts to
recentralize in the 1980s by a revitalized LDP and PM Nakasone although many
alliances of ministry-politicians-interest groups held a lot of autonomy.
As it happens, I criticized this article pretty sharply in a piece that never
quite got finished enough for publication called "Fragmentation and Power." I
thought that the picture Pempel and others drew of a smoothly coordinated
conservative camp (Japan, Inc) in the 1960s was not accurate, and indeed that
research by American political scientists carried out at the time (Thayer,
myself, Pempel and Fukui themselves) pretty clearly showed that things were
pretty pluralistic and conflictual back then. But while Pempel is more than able
to defend his own arguments, let me note here that the factors he was analyzing
are a good deal more productive in understanding change in policy and
policymaking than are issues of ultimate or formal authority.
That is, Pempel's argument was that until the early or mid-70s, the LDP, the
economic ministries and big business were so much in agreement over major policy
goals that everything was managed smoothly. In that case it doesn't matter who
has the final authority. Actually he says the bureaucracies had influence and
took the initiative then, but he doesn't say they are in control and indeed
there is nothing here to contradict a picture of delegation by the party. In the
1970s it all fragmented and everybody started squabbling; he actually deos not
mention anything about party VS. bureaucracy here, but he does say that party
politicians got more active and knowledgeable and therefore more influential
over the content of policy, as did local governments, courts, a variety of
interest groups, and the opposition parties.
Now I think that activation of party politicians and a resulting rise in
influence in policy making (esp. budgeting) occurred much earlier, but I know
that this is the right factor to look at--how do politicians' motives to
intervene, necessary knowledge, and power resources in terms of alliances with
interest groups etc change, and affect what they do in policy making? (And one
can say the same for ministries, by the way.)
As I said a day or two ago, our selection of which independent variables explain
the most depends mainly on the dependent variable that interests us, what we
want to explain. Pempel and I want to explain policy, what the government does,
so we are interested in how problems get defined and where ideas for solutions
are coming from, as then of course what happens to them during the process (e.g.
whether there are real amendments in the Diet--no in the 60s, yes in the 70s, as
Pempel emphasizes).
The question of whether or not there was a "power shift" (and in fact Pempel did
not assert that there was) is not too relevant here, and neither is most of the
analysis I have seen written from the P-A framework.
Although now that I think of it, Thies himself wrote a piece arguing that the
government's agriculture policy changed abruptly when rural Dietmen lost their
majority in the LDP, in 1978 as I recall. The argument in substance was the same
as Curtis (Japanese Way of Politics) and Pempel (this same piece and also
Creative Conservatism) had already made (I don't think that point was cited in
the article though). That is, the LDP could no longer depend on its old rural
(and small business) base and tried to find new constituencies with new
policies. But Curtis and Pempel saw this process as a complicated evolution
while Thies (and McCubbins I think) saw it as a quick simple turnaround. The
latter is much more elegant, but in my view when one looked at the details of
both the independent and dependent variable measurement it doesn't hold up.
Actually I don't agree with Curtis and Pempel either, but let me not get into
that. To return to the original point: Thies brought up Pempel because John
Campbell asked for a cite to works that conceive the relationship between burs
and pols as a power struggle with, to quote myself loosely, "no identifiable
resources or rules..."

The piece he cited does not see a power struggle between burs & pols, and it
actually is pretty precise about resources--although they are a bit complicated.
It doesn't need to say much about rules because it isn't about rules, it is
about policy.