Theoretical Foundations

Legislators have preferences about policy decisions. These preferences are based on their partisan slant on the issues, on the degree to which they wish to be representative of their constituents' desires, on their responsiveness to organized interests, and on their personal views about politics and good government. On any particular issue, politicians will take a wide variety of positions, based on preferences ranging from very liberal to very conservative. Looking at an issue, we often find the current (status quo) policy somewhere near the middle of these preferences, not as liberal as many Democrats would like it, and not as conservative as many Republicans would wish. This much is obvious, and results largely from legislative compromise.

Much more can be said about when bills will be passed and what the outcomes will be on a liberal‑conservative scale. The theory on which this book focuses is introduced and analyzed formally in Krehbiel's "Institutional and Partisan Sources of Gridlock: A Theory of Divided and Unified Government" (1996). Krehbiel's Pivotal Politics: A Theory of U.S. Lawmaking (1997) includes a relatively non-technical overview of the "pivotal politics theory" and subjects the theory to a wide range of empirical tests on legislative action that occurred throughout the postwar period. As a general matter, we regard his studies and ours to be complementary with respect to the tests, analyses, and ‑interpretations. Throughout this chapter we will note differences between Krehbiel's formal analysis and our portrayal with regard to uncertainty, exogenous shocks, and bargaining between the President and Congress over final outcomes

The "revolving gridlock theory" is based on a one‑dimensional spatial model.4 We claim that, on any particular issue, legislators can be assigned positions from the most liberal to the most conservative. As a practical matter, preferences may be revealed through interest groups' ratings or other measures of legislators' positions on a variety of issues.5 We explore these practical issues in greater detail in Chapter 3 with regard to the 1980 electoral shift, in Chapter 4 and the Appendix with regard to preferences over time, in Chapter 5 with regard to interest group ratings, and in Chapter 6 with an example of one particular representative.

In addition to the positions of individual legislators, the position of the status quo policy on each given issue likewise can be discerned. Based on the position of that status quo point with respect to the positions of the members of Congress, we can speculate with some accuracy where a bill will need to be positioned to pass successfully through the institutional structure of lawmaking. The Senate filibuster and the presidential veto provide the institutional constraints on policy‑making according to the revolving gridlock theory. If a bill is to become law, it must gain a majority in both houses and must not be killed by a filibuster or a veto. We argue that these constraints caused by legislators' positions and supermajority institutions are the reason policy gridlock is prevalent in the American legislative arena today; and a change in the party of the President is not sufficient to bring about an end to this gridlock.

The first institutional feature of note is the filibuster. The Senate has always been known for its slow and deliberate consideration of issues. In particular, a Senator, once given the floor, can continue to speak for extended periods of time. When a Senator's right to hold ' the floor indefinitely is utilized to slow or stop the advancement of a bill, the action is commonly referred to as a filibuster. The filibuster gained particular notoriety during the passage of civil rights bills in the 1950s and 1960s. In one instance, Strom Thurmond of South Carolina, speaking out against civil rights legislation, held the floor for twenty‑four hours and eighteen minutes. Obviously, filibusters could keep the Senate from acting on important legislation. As a result, the Senate has over time adopted rules limiting the use of the filibuster. Of great significance is Senate Rule XXII, allowing for a cloture vote to end debate. To invoke cloture, sixty Senators must agree that the issue has been sufficiently discussed and that the Senate should continue on with its business, often leading to a vote on the bill being filibustered. The cloture rule thus limits the power of any small group of Senators who wish to talk an issue to death. But it still allows a minority to have significant power over an issue. If forty‑one Senators wish to kill a bill through a filibuster, they can do so by voting against cloture. This institutional feature thus can have a great impact on policy outcomes.6

Figure 2.1 helps illustrate this point. The range from FL to FR represents the central twenty members of the Senate, with M being the median member. The forty-one Senators to the left of and including FL could successfully filibuster a bill. Likewise, FR and the forty Senators to the right could successfully filibuster. The Senators are placed along this line based on their positions on the issue at hand. For example, if we are looking at minimum wage legislation, legislators could be lined up based on what dollar level they feel is appropriate for a minimum wage.7 If the status quo (Q) on a particular issue is between FL and F', we argue that no R policy movement can occur. That is, if the central twenty Senators are arguing that the minimum wage should be between $4.00 and $5.00, and the current minimum wage is $4.25, that wage cannot be changed. Looking again at Figure 2. 1, if the majority to the right of Q attempts to enact legislation moving policy to the right, FL and the 40 Senators to the left will filibuster to prevent any legislative movement. This does not mean that the minority on the left can dictate policy, however. Indeed, if they attempt to move policy any further to the left, FR and the forty Senators to the right will filibuster to prevent that movement. Thus policy Q cannot be changed by the Senate. This analysis holds true for any status quo policy in the range between FL and FR. Because, in this example, a majority would like to enact a more conservative policy but no change is possible, the institutional feature of the filibuster alone is enough to lead to cries of "gridlock."

This "gridlock region" within which no policy change can occur is actually even larger than described above. The reason for this is found in a second institutional feature: the presidential veto. If the President adopts a conservative position on an issue, the region of inaction is extended further to the right. The logic here is much the same as with the filibuster. If the status quo policy is fairly conservative and Congress acts to make the policy more moderate, the President can veto that legislation. Instead of needing the forty‑one conservative Senators required to maintain a filibuster, the President only needs thirty‑four conservatives to sustain a veto.

| Gridlock |

Liberal Preferences-------F_L—Q—M----F_R----------Conservative Preferences

FIGURE 2.1 Policy Constraints Caused by Filibusters

Because a cloture vote requires three‑fifths of the Senate and a veto override requires two‑thirds, the veto provides a greater constraint on policy action. When the President is conservative and the Senators are ranked along the main policy dimension, then, this region of inaction, or gridlock, stretches from the forty‑first Senator to the sixty‑seventh. With a liberal President holding veto power, this region stretches more to the left, from the thirty‑fourth Senator to the sixtieth. If previous policy has positioned the status quo in this region, then Congress can successfully undertake no further policy action. Movement to the left or the right will be halted by successful filibusters or vetoes, as indicated by points FL and V in Figure 2.2.

In these diagrams, we are defining the edge of the gridlock region nearest the President to be determined by the number of legislators, based on their preferences, necessary to override a veto. However, if a President is in a position more centrist than this veto‑pivotal member (denoted as "Y' in the figure), the constraint will then be the President and not this member of Congress. As such, compromises will have to appeal to the President for the sake of avoiding a veto, rather than appealing to enough members of Congress to override the veto. Thus the gridlock region could stretch from the filibuster point to the veto point or to the President's ideal policy point. For extremely centrist Presidents, the veto is of little concern and the gridlock region is defined only by the filibusters, as in Figure 2. 1.

After Congress passed the minimum wage increase in 1977, the new wage was securely in the region between the thirty‑fourth and the sixtieth Senator. A movement to increase or decrease the wage could be stopped through either a filibuster or a veto. Indeed, even when President Carter asked his own party to delay the minimum wage increase, there was not enough support for congressional action. From 1977 to 1989, policy gridlock reigned on the minimum wage issue. Very liberal Congressmen again saw the wage slip out of line with their own preferences, and they noted that they could not change policy. They were frustrated with what they called partisan gridlock. Very conservative members, on the other hand, saw the declining value of the minimum wage as more in line with their policy preferences. They had no reason to complain about policy inaction. Eventually, however, inflation and other economic conditions caused the purchasing power of the $3.35 wage to fall so far behind that it again became out of line with the preferences of the majority of Congress. At that point, gridlock was brought to an end under a Republican President.

| Gridlock |

Liberal --------------F_L----Q--------M--------------V------------------Conservative

FIGURE 2.2 The Full Gridlock Region

The assumption that the extensive use of filibusters and vetoes is favored by Senators and Presidents is questionable. Is it actually in the interest of a President or of a group of Senators to repeatedly veto or filibuster legislation? Would the political costs associated with being labeled an "obstructionist" not outweigh the policy benefits? George Bush's repeated vetoes perhaps even helped lead to his electoral defeat in 1992. But his defeat was not the necessary outcome of his vetoes. With an aggressive campaign, he could have used the vetoes to argue that he was fighting against the liberals in Congress. Indeed, this was precisely what he argued at the end of his campaign. As it turned out, the press had already characterized him as a do‑nothing‑at‑home President. And, come election time, the public had bought this story.' And as for the conservatives filibustering early Clinton policies in the Senate, they typically represented constituents who were pleased to hear that liberal policies were being defeated. Although conservative Senators may have preferred legislative action on many issues, stopping action that would have been against their constituents' interests was considered good work. Nevertheless, it is important to be aware of how filibusters and vetoes are perceive an whether this affects how often they are used.9

The gridlock region described above is important with regard to policy action as well as policy inaction. Figure 2.3 shows the policy region for the Senate with a conservative President, stretching from the filibuster point (F) to the veto point (V). The Senator at point F plays a pivotal role in policy formation. If the status quo policy is to the right, that Senator joins the forty liberals to the left in filibustering any further movement to the right. If policy is to the left, then the pivotal Senator allows a shift to the right just so far as is in that Senator's interest. The pivotal Senator will join the forty colleagues to the left to filibuster bills that go too far. We refer to this Senator as the filibuster pivot, as this lawmaker plays a pivotal role in deciding which bills are satisfactory and which should be filibustered.10 The Senator at point V holds similar powers concerning policy shifts to the left, and is referred to as the veto pivot. The thirty‑three Senators to the right can be joined by the veto pivot to sustain a presidential veto. Likewise, the sixty-six Senators to the left can be joined to override a veto. Thus this Senator's position is pivotal in deciding whether a bill is conservative enough to pass the Senate, even with a veto threat. We call the region between the filibuster pivot and the veto pivot the gridlock region. Policies in this region are maintained, whereas those outside are moved inside.

|||||||||||||||||||||||||||||||||

Q--------F--- Q'---M------------V

FIGURE 2.3 Possible Outcomes with a Filibuster Threat

Looking again at Figure 2.3, the point Q represents a status quo that is outside of the gridlock region. The Senate can thus take successful action in this policy area; but this action is again constrained by the threat of a filibuster. If the Senate proposes a bill that would shift policy to the right of Q'I the bill would certainly be filibustered. The Senator at F and the forty Senators to the left all would be disadvantaged by a bill to the right of Q because such a bill is further from their ideal policy than was Q. If a bill is proposed that would move policy to Q') the Senator at F is indifferent about whether to let the bill go through or to filibuster to stop it. Because of the possibility of a filibuster, the only policies that can be adopted are those in the shaded region between Q and Q In actuality~ the predicted outcome is somewhere between F and Q'I 1 All movements from Q toward F are advantageous to the Senator at F. Only as a proposal becomes more conservative than Senator F's ideal policy does this Senator consider a filibuster to halt bill movement toward the right. The policy will end up between F and Q'; the exact position of the bill in this region is subject to agenda setting and political bargaining." The conservative President and a majority of Senators prefer the point Q' ' but the filibustering group, often less impatient than the majority to get something passed, prefers point F. The minority might filibuster until a compromise is made, or a few Senators might find their position unpopular back home, causing the filibuster to break and the majority on the right to benefit.13

A diagram similar to Figure 2.3 can be drawn with regard to minimum wage legislation in 1989.14 In Figure 2.4, the status quo policy has minimum wage at $3.35 an hour, as it was when Bush took office. The veto pivot is positioned at $4.00 an hour.15 The region between $3.35 and $4.65 is shaded, showing policy outcomes that would be advantageous to the legislator at the veto pivot. As legislation allows the minimum wage to rise toward the $4.00 mark, the legislator preferring the $4.00 wage will vote for it over the status quo wage. That legislator would even prefer a wage at $4.50 to one at $3.35. But when the proposed minimum wage reaches $4.65, the veto pivot legislator becomes indifferent; a wage that is 65 cents too low (i.e., the status quo) and one that is 65 cents too high (i.e., the proposed new wage) are equally unpalatable. If the proposed wage is above $4.65, the legislator at the veto pivot ($4.00) and those who are more conservative will oppose the change. The bill will be vetoed by the President and that veto will be sustained.

It should be clear that the outcome of this legislative process will be a minimum wage set between $4.00 and $4.65. The vast majority, including the veto pivot, prefer a raise to at least $4.00. And the veto pivot is indifferent to a choice between the status quo wage and a $4.65 wage. Liberal and moderate members of Congress would attempt to get as close to the $4.65 wage as possible, whereas conservatives would try to keep the wage increase as small as they could. At this point there is a standoff.

Minimum Wage Levels $3.35 $4.00 $4.65

|||||||||||||||||||||||||

Q V Q' M F

FIGURE 2.4 Possible Outcomes on Minimum Wage Policy (1989)

The initial Democratic proposal in 1989 was a minimum wage of $4.65. Bush needed to establish a counter position. If he picked a position that increased the wage too much, ‑his position would be frowned upon by his party and his electoral constituency. If he picked too low a wage, his coalition on the right could be broken by a compromise proposal. Some of the legislators he was counting on to sustain a veto could be won over by the majority with the argument that Bush was appearing too conservative and too much a proponent of gridlock. Bush knew that some sort of a wage increase would pass. The size of the increase would depend on the President's position and on the political bargaining game.

In short, the revolving gridlock theory predicts that status quo policies inside the gridlock region will be maintained, and policies outside the region will be brought inside, usually through minor policy adjustments. With very little resulting policy change, even where a majority approves change, this model starts to explain what is referred to as policy gridlock. When this gridlock occurs under unified party control of government, we call it unified gridlock. Unified gridlock resulted under the 103rd Congress during the first two years of the Clinton administration.

It should not be assumed that this model rests on the observance of filibusters and vetoes. The mere threat of a veto or a filibuster is often enough to kill a bill or to force it to be altered so as to override a veto or to gain sufficient votes for cloture. Successful vetoes and filibusters might actually be quite rare. Because time and effort are scarce commodities in Congress, it would be easier for the majority and the leadership to abandon a bill early on than to lose it to a filibuster or veto. However, in some circumstances politicians may wish to go down swinging.16 Opponents can raise the issue of repeated sustained vetoes, such as those in the Bush presidency, as an example of the President and his party causing gridlock. And Democrats could claim that the repeated filibusters by Bob Dole and other conservatives in the 103rd Congress were "obstructionist. ' Of course, on the other side of the coin, during the 1994 elections Republicans effectively claimed that the Democrats were poor at policy making, unable to pass major legislation even with control of Congress and the presidency.

The above discussion has concentrated mainly on the Senate. There similarly exists a gridlock region for the House. As filibusters are not allowed in the House, this region only stretches from the House median to the House veto pivot. With a conservative President, status quo policies in this region cannot be shifted to the right because a majority would not vote for such a shift, and policies cannot be moved to the left because such a shift would be vetoed and the veto sustained.

105th House

V_H M_H

|||||||||||||||||||||||||||||

V_S M_S F

105th Senate

FIGURE 2.5 The Gridlock Region in the Bicameral 105th Congress

Because this region is smaller than in the Senate, it is often less of a constraint on policy.17 However, the need for a supermajority to override a veto is a serious constraint in both the House and the Senate.

Figure 2.5 shows the bicameral situation as we perceive it today, with a fairly conservative House and Senate and a Democratic President. Here the gridlock region is defined by the constraints of the filibuster pivot in the Senate and the veto pivot in both the House and the Senate. If the House were considered more conservative than the Senate (as was the case immediately following the 1994 elections), the gridlock region in the figure would expand out to the further of the veto pivots‑that of the Senate. Indeed, as we argue in Chapter 6, the constraints on policy change as outlined by the Contract with America during the 104th Congress were the presidential veto and the preferences of the veto pivot members in the Senate. The gridlock region for our bicameral system is determined on the left by the greatest restriction between the two houses on movement to the right, and is determined on the right by the greatest restriction on movement to the left. This usually means that one of the restrictions is the filibuster in the Senate, and the other is whichever veto pivot extends the gridlock region further away from the median.

The position of the President and his veto threat determines which side of the gridlock region will be stretched out to the veto pivot. With the election of Bill Clinton, the previous extension of the gridlock region to the veto pivot on the right was altered such that the filibuster pivot is now on the right and the veto pivot is on the left. Policies that were held in place by Bush vetoes were released for action by Congress. But they were still constrained by the filibuster pivot in the Senate. Unless President Clinton's proposals were made suitably conservative, the Senator at the filibuster pivot and colleagues to the right would have voted to stop the legislation with a filibuster, voting down cloture attempts until the bill died or a compromise was reached. This was the fate of many Clinton proposals in 1993, including the national service legislation and the jobs stimulus package.

Following the 1994 elections, policy proposals reflected the shift in preferences to the right, advancing more conservative policies. The limitation on these policies was the threat and use of vetoes by President Clinton. The President's successful vetoes of Republican budgets and welfare overhauls made Congress adjust their proposals more to the President's liking. Had Bob Dole been elected in 1996, the constraint on policy change to the right would have been the less‑restrictive filibuster in the Senate. Many items in the Contract with America that were watered down to avoid or override Clinton vetoes could have been enacted in full.

Uncertainty

"It's Friday. Some people are confused," Bob Dole explained. The Friday in question was September 23, 1988, and the Minority Leader had just lost five more Republican votes in his attempt to keep the minimum wage filibuster going. The cloture vote had still failed, as nine Senators had not been present to vote. But Dole was concerned about the apparent weakening of his coalition. This issue was a bit complex, as they were voting on a procedural issue, not on whether to actually pass a higher minimum wage. But wasn't it a straightforward procedural vote? Maybe not.

The actual impact of an increase in the minimum wage was uncertain. Earlier in the summer, the San Francisco Federal Reserve reported that a minimum wage increase could boost inflation by a noticeable amount. Since then, several studies were released claiming that the proposed minimum wage hike could cause the loss of up to a million jobs. Such news raised questions among the Senators voting on that late September day. How soon would these effects on the economy take place? Would the voting public be able to trace the blame to those who voted for an increase in the minimum wage? Would these hardships counterbalance the benefits that constituents would perceive from increased wages for those with low incomes?

The actions of interest groups raised more questions in the minds of legislators. Even if no economic hardships materialized before the elections, would the varied interested parties influence the elections depending on my vote? If I vote for a wage increase, will my support for small businesses diminish? If I vote against an increase, will organized labor raise more support for my opponent? Even Dole was frustrated with the involvement of some of these interest groups. "How many Labor Committee bills from organized labor do we need before we go home?" he asked.

But Dole had not taken steps to make the issue clearer for his colleagues in the Senate. In fact, he in muddied the waters further, by tying action on minimum wage to the approval of Reagan's nominees to federal judgeships. "I am advised by members on my side to bring this place to a halt until we get some action" on the judicial appointments, Dole said. This led to even more questions. If a Senator supported the filibuster, would he then be perceived as being involved in a type of blackmail to get conservative judgeships?

With the issue so complicated and the uncertain prospects of elections so close on the horizon, Dole should not have been surprised to see some flip‑flops in voting. Over the weekend he would make sure that he would not lose the next cloture vote, leaving the Democratic leadership with the lose‑lose proposition of either prolonging the stalemate or dropping the bill. By Monday it was clear that the conservative filibuster could continue indefinitely, so the Democratic leadership dropped the bill.

In forming policy, legislators face uncertainty on a number of levels. Of concern to the theory here are two types of uncertainty: uncertainty over the actual policy results of passing a bill, and uncertainty over constituent reactions to voting for or against a bill." Although members of Congress are forced to live with some level of uncertainty, they take many steps to minimize that uncertainty.19 They listen carefully to constituents, paying attention to surveys and polls. They take advice from experts, whether committee members who have specialized in a policy area or authorities who give testimony in hearings.20 Still, the remaining uncertainty leads to mistakes.

First, there is uncertainty over the actual policy results of passing a bill. In the above section, we assumed that the status quo policy and the alternative proposed by the bill were known and were easily placed on a one‑dimensional line. Legislators then simply pick whichever policy is closer to their preferred outcome. In reality, policyrnaking is an uncertain activity. Budget estimates made over a five-year period are undoubtedly going to lose accuracy over time. Members of Congress cannot accurately predict which interpretations and actions government agencies and bureaus will take. Policy- makers and policy analysts are unsure of just how many people will qualify for programs, find loopholes, or be indirectly affected by a policy change. Legislators try to find out as much as they can about the policy consequences of various actions, and then they must take a risk and vote.

The uncertainty of policy outcomes could have a beneficial or adverse effect on the chances of a bill's passage, depending on perceptions of the status quo. When the public seems pleased with present policies, it may be difficult to pass new legislation. Although the proposed policy could actually improve the status quo, legislators feel no pressure to take chances. Such was the case with minimum wage legislation in 1988. Pollsters predicted that incumbents would fare quite well in that year's elections. The economy was fairly strong, and it appeared that George Bush would be elected President. Congress felt that tinkering with the minimum wage while otherwise in such a strong position so dose to the elections raised too much uncertainty. The San Francisco Federal Reserve reported that a minimum wage increase could boost inflation by a noticeable amount.21 Analysts released a variety of studies claiming that a minimum wage hike could cause a loss of up to a million low‑wage jobs. This potential job loss combined with the grassroots lobbying campaign by business groups led the House leadership to avoid votes on the bills proposed out of committee.22 Although Congress did consider the issue, there was no outrage over legislative inaction, and if Congress had acted there was a great risk of making economic and political conditions worse.

In contrast to this case of uncertainty leading to inaction, at times politicians embrace an uncertain outcome. When the current policy is considered poor, the public perceives that things cannot get any worse. This paves the way for a string of new, uncertain policy proposals. Had economic conditions been different in 1988, perhaps members of Congress would have supported a minimum wage increase. For example, if there were an outcry about large numbers of working Americans living below the poverty line, legislators could address the issue through a minimum wage hike just before the election. The public would think that Congress was being responsive, and the inflation and job loss would not be felt before the election. Whereas typically members of Congress are looking to avoid uncertain options, under these conditions they may embrace a major change.

Thus uncertainty over policy outcomes can either add to or relieve policy gridlock. When legislators are uncertain about the consequences of their actions, they may either take small steps or make no changes to current policy. But when there is demand for immediate action, uncertainty could trigger a major policy change that would have been unacceptable to legislators had they been perfectly aware of the outcome. As the uncertainties are resolved over time, members of Congress can better judge what they have done. If the results are found to be in the "gridlock region,' no further policy changes are made. Such was the case with minimum wage policy under President Carter. When he realized that the wage increase he had called for was hurting the economy, Carter asked Congress to delay its implementation.23 However, so many members of Congress agreed with the wage increase that it went forward as planned. Even though the policy outcome was more dramatic than expected, the results were still in the gridlock region. If the uncertainty is resolved with the discovery that the policy enacted did not move into the gridlock region at all, or even that it overshot the region, modifications to the policy will be made. For example, when the 1981 tax cuts in concert with the economic downturn yielded large budget deficits, President Reagan supported minor tax increases over the next few years. And when the catastrophic health care measure passed by the I 001h Congress was found to be catastrophic itself, it was quickly repealed.

In addition to being uncertain about where the policy outcome of a bill will lie, members of Congress face a second uncertainty: how their constituents will react to how they vote. In the above section, we argued that members of Congress are aligned from liberal to conservative. Their positions on various issues can be determined by observing how they vote over time. When they vote, members of Congress seeking reelection must be aware of how their constituencies view their votes on the issue at hand. And yet these members are uncertain as to what the reaction will be back home. Many policy votes will simply be ignored by constituents; others will be observed but play little or no role in swaying voters; and still others will become major campaign issues. Furthermore, the uncertain reaction of constituents is compounded by the uncertain policy outcomes. Even though they may have the best of intentions, legislators will be blamed for the unforeseen results of their actions (or inaction).

Because legislators are trying to maintain their popularity among constituents, they must try to resolve the uncertainties concerning voters' perceptions of the issues. In particular, members of Congress attempt to reduce uncertainty by judging the salience of an issue to the public and the leanings of the public either for or against a bill. Legislators pay attention to surveys and polls as well as to calls and letters from constituents. Over time, they settle on a position with which they are comfortable, whether they derive benefit from their party, from their constituents, from funding via Political Action Committee (PAC) support, or from a sense of doing what is right for the American people.14 Throughout 1988 and early 1989, lawmakers were deciding on their positions regarding minimum wage increases. They were careful in 1988 not to upset the labor movement on one side of the issue or business groups on the other. Either side could have helped finance challengers against incumbents who weren't carefully positioned.

As the legislators themselves are converging on their final positions on a particular issue, the bill being produced is also changing. Support for a legislative change in policy early in the formulation of a bill does not necessarily mean that anything will be done in the end. It certainly does not mean that the President or the majority party is free to push through its bill of choice. As uncertainties are resolved about where the status quo, the various new bills, and the legislators' own positions lie, we can fairly accurately predict the policy outcome. If the status quo is found to be in the gridlock region with a large number of legislators on each side of it, policy change can be stopped through filibusters and vetoes.26 If the status quo is found to be outside the region, as was shown in Figure 2.3, it win be brought inside the region, but seldom as far as those promoting change had wished.

On many issues Congress deals with, most of the major uncertainty can be resolved before action is taken. In such cases, the revolving gridlock theory should be expected to hold without exception. Additionally, we can accurately predict when uncertainty will play a large role. Uncertainty influences policy when legislators face two constraints: complexity and time. When dealing with a multiyear budget and an uncertain constituent reaction, or with bills involving the one‑seventh of the American economy that is health care today, legislators can make mistakes. The issues are so complex that all of the uncertainty cannot be removed. Also, when dealing with the time constraint of enacting a policy just before elections, uncertainties cannot be removed from the legislative process because all the variables cannot properly be addressed. Although members of Congress wish to campaign on the strength of a new program they have just passed, they do so at the risk of finding that they have made a mistake when they return to Congress the next year. We claim that the revolving gridlock theory will hold true generally, even where uncertainty surrounding legislative decisions makes the process more complex.

Elections and Exogenous Shocks

"I can't tell you what a very significant departure this is from what we've had over the past eight years." Democratic Senator Ted Kennedy was obviously pleased that minimum wage was again being discussed in March 1989. He had seen the wage shrink in terms of purchasing power for years, and had been a strong supporter of an increase. To raise the minimum wage to $4.60 an hour would only restore the purchasing power it enjoyed in 1981, Kennedy argued. It would not make up for the years of decline that people living on minimum wage had endured.

Now was the time to do something about the minimum wage. The newly elected President had issued a campaign promise to raise the minimum wage. And he had not brought in many conservative members of Congress on his coattails who would oppose a wage hike. As such, a large majority in both houses of Congress supported an increase.

Although President Bush had only supported an increase to $4.25 an hour, Kennedy believed that a greater increase would have enough support to override a veto. But would Bush be foolish enough to veto the increase? About 80percent of the public favored an increase in the minimum wage. If Bush vetoed the bill passed by Congress, he would lose credit for any increase whatsoever, and would hopefully lose a veto fight as well. Kennedy therefore wanted to propose an increase that would restore much of the purchasing power of former wages, and one that would also be an embarrassment to Bush if he vetoed it. On March 23, the Senate decided on a wage of $4.55, just three thin dimes more than Bush's proposal. With the approval of the House, the ball would soon be back in the President's court.

In previous pages we defined the gridlock region and argued that policies in that region could not be changed, due to the diverse preferences of members of Congress and to the institutional structure involving the use of filibusters and vetoes. Status quo policies outside the gridlock region are brought just inside it. Because of uncertainty, policies are sometimes formulated that are outside the gridlock region. We claim that when these outlying policies are discovered, they are brought back inside the gridlock region. If all of this is true, it would seem that over time Congress would get around to dealing with every policy area, bringing each policy outcome into the gridlock region where it is then held in place. We argue that this would be the case if there were no shocks to the system. But in the constantly changing world of politics and policy, a number of exogenous factors impact upon the theory. The exogenous factors of greatest concern to us are threefold: the election of new politicians, the changing preferences of constituents, and the shifting of policy realizations over time.

Because the narrow band that we call the gridlock region is defined by the preferences of legislators in the House and Senate, a change in these members or in their preferences could alter the size and placement of the region. If such a change leads to a state in which some current policies are no longer within the region, then those policy areas are released for legislative action. The most straightforward way to change the placement of the gridlock region is to change the members who define its position. For example, if the Republicans gain a number of seats in a given election, the region is likely to shift to the right. This may result in no policy change, if there were no fairly liberal status quo policies just inside the gridlock region. Or it could result in across‑the‑board changes, if a large number of status quo policies are released from the gridlock region.

Thus policy gridlock depends on both the size and the shifting of the gridlock region. The size of the region is determined by the difference in preferences between the pivotal members on both edges of the region. If the preferences of the forty‑first (filibuster pivot) and sixty‑seventh (veto pivot) members are quite similar (as with a large number of centrists), little gridlock will occur. However, if the preferences of these members are quite different, policy changes over a vast range of policies will be unachievable. This is perhaps why commentators worry about the "lack of a political center" in today's Congress. In the Appendix, we show the distribution of preferences in Congress since 1980. Note that, given these preferences, the gridlock region has been sizable in recent years. However, not only the size of the gridlock region but also its movement matters. Even if the gridlock region had about the same size after the 1980 elections as before, the dramatic shift to the right released a number of policies that were formerly in the gridlock region and that could now move to better reflect the preferences of Congress and President Reagan.21

However, most elections do not bring about sweeping changes. The vast majority of incumbents are reelected.29 Additionally, changes at the margins often do riot affect the position of the gridlock region to any great extent. If ten liberal members and ten conservative members all lose their seats and are replaced by new members on the opposite sides, the region may not be changed at all. The moderate legislators at the endpoints of the gridlock region may remain the same, whereas the legislators on the wings change dramatically. Furthermore, voters may take into account the preferences of the other members of Congress and of the President in order to strike some sort of balance or include checks on other politicians. Although we are indeed interested in the link between policies and elections, our main focus is on what happens when politicians are already in the government, emphasizing that it takes major electoral shifts to move policy left or right.

If the gridlock region does move noticeably in one direction, this is often the result of the "coattails" effect frequently found in presidential elections. If the party of the President changes from Democrat to Republican and a number of conservative Senators and Representatives are brought in with the new President, liberal status quo policies become vulnerable to change. The President win certainly wish to enact change in all of those policy areas before the typical loss of seats during the midterm elections.32 This explains the phenomenon known as the "honeymoon" in which a new President is able to quickly bring about policy change on a number of issues.13 (Often within the first few months of a presidency, most of the policies with low uncertainty are brought back into the new gridlock region.) Because policy movement is still restricted by the possibility of a filibuster in the Senate, the President often will not get all that he wants during his honeymoon. The more extreme a President's position, the more he will appear to be losing in the compromises he is forced to make. Because the policy outcomes illustrated by the above figures were not dependent on how extreme the president's position was, a President often appears more powerful by asking for what he knows can be passed rather than by asking for what he wants and then having to abandon the policy or cave in to the opposition.34

This is an important point in our understanding of presidential requests. It is possible that President Bush wanted the minimum wage to stay where it was. But this was an unpopular position in Congress, where a raise to at least $4.00 was inevitable. By taking a position supporting a raise to $4.25, Bush appeared to be moderate while also challenging those who wanted an increase beyond $4.25. Had he positioned his proposal any lower, he may have appeared too conservative, risking the breakdown of his coalition. Had he requested a higher wage, Bush would have been giving away more than he had to, perhaps hurting the economy as an outcome. Had he ignored the issue entirely and left Congress to pass increases that he would veto, Bush would eventually have had to give in to a small increase or risk having his veto overridden and the steam taken out of his presidency early in his first term. Bush proposal turned a losing issue into an early victory.

After the honeymoon, the President is left to deal with those policy areas containing the greatest uncertainty, those policies that cannot easily be adjusted back into the gridlock region. This is when the legislative challenge really begins. President Reagan was able to use the uncertainty surrounding supply‑side economics to his advantage. President Clinton, on the other hand, lost popularity in his first year by trying to enact his own ideal policies only to end up giving concessions to conservative Senators. In 1994, when facing the uncertainties of health care reform, members of Congress tried to distance themselves from Clinton's unpopularity.

The gridlock region can be changed not only through legislator turnover but also through a shift in the preferences of legislators. As described above, legislators' preferences are complex; they depend on party strength, on public opinion of their constituencies, on the influence of interest groups, on the amount of available information, and on the legislators' own personal preferences independent of outside forces. When any of these factors shifts dramatically, a legislator's preferences will follow.

If the President is enormously popular and has agreed to campaign for members of his party who support him, those legislators are able to vote with the President even if they lose some support from interest groups or constituents. Any such loss of support can be won back in a grand fashion if the President campaigns on behalf of the legislator. But a President without such popularity will see members of his own party distancing themselves from him. Likewise, if public opinion is shifted through advertising campaigns, such as the "Harry and Louise" ads against the Clinton health care plan, legislators may alter their voting preferences in order to please their constituents. And if interest groups join in on an issue at the last minute, or provide new information to Congress, a shift in legislator preferences is again likely.

When these shifts are dramatic, members of Congress may change their preferences to the extent that policies previously in the gridlock region are now to the left or right of it. In such cases, Congress will pass bills to bring policies back in line with the preferences of its members. Of note here is a beneficial side effect of gridlock. Because of the lack of policy movement in the gridlock region, members of Congress do not spend all of their time chasing small shifts in public opinion. If it weren't for the threat of filibusters and vetoes, legislation could be raised and passed with the slightest change in legislator preferences. These institutional constraints sometimes mean that the majority's proposals can be stopped by a minority, but they also encourage Congress to act only when significant action is needed.

The first two exogenous shocks to the revolving gridlock model‑a change of legislators and a change of legislators' preferences‑affected the positioning of members of Congress, and thus of the gridlock region. A third exogenous factor involves the position of the status quo policy relative to the gridlock region. As was discussed above, there is often uncertainty with regard to the actual outcome of legislation adopted by Congress. Although some laws have been unchanged on the books for decades, others are changed all the time. Congress is unable to specify a budget for more than a year or two into the future. Minimum wage, Medicare, and Social Security benefits are constantly being adjusted. With an increase in crime, a crisis overseas, a natural disaster, a dramatic increase in the cost of health care, a newly discovered disease, an economic recession or depression, or any of a number of other unforeseen events, new issues constantly arise for Congress to deal with.

When new issues arise, or old issues take on a new significance, members of Congress face a high level of uncertainty as to what should be done. Congress devotes most of its attention to dealing with these unpredictable issues when they arise, then returning to the normal business of adjusting policies that have drifted outside of the gridlock region. With the economic difficulties of the late 1970s and early 1980s, Congress dealt with the problem through an attempt at supply side economics. When this reaction led to growing budget deficits, Congress became seriously restricted in its ability to initiate new programs and was forced to raise taxes and limit spending just to avoid a crisis.

When President Bush reacted to lr4s invasion of Kuwait, Congress had to address the possible use of military force if the legislative branch wished to maintain a reasonable level of influence in the foreign policy arena. When hurricanes hit Florida and Hawaii and floodwaters covered the Midwest, again Congress had to act. Shocks to minimum wage policy were slower in coming. Inflation during the 1980s caused wages to lose about one quarter of their purchasing power. When the minimum wage had thus dropped to levels unacceptable to legislators, they called for a wage hike. This was not the result of a liberal shift in the preferences of members of Congress, nor in divided or unified government. Outside factors had caused the minimum wage to become skewed away from the preferences of lawmakers, who therefore changed the policy

Most of the significant action taken by Congress has occurred as a reaction to exogenous shocks to the legislative system. When new legislators are elected and there is a major shift to the left or right, new legislative action is likely. When public opinion solidifies behind an issue, such as during the Great Depression, the world wars, or the space race, the preferences of legislators are altered enough to bring about significant change. And when unforeseen policy shocks arise, legislative action is again possible. Often these shocks occur under extraordinary circumstances. The revolving gridlock theory is significant in its ability to explain the results of these occurrences as well as to address the day‑to‑day legislation passing through Congress. When lawmakers fail to respond to exogenous shocks, they put their seats at risk. The inability of Republicans to react to the events of the Great Depression led to enormous Democratic victories. When the Democrats failed to respond to cries for tax relief throughout the 1970s, Republicans gained control of the Senate and presidency and found enough support to enact huge tax cuts in 1981.

In the absence of major electoral and policy shocks, the government falls into periods of legislative gridlock during which either no policy is passed or only incremental changes around the edges of the gridlock region are made. It is our contention that the election of Bill Clinton with only 42 percent of the vote and few coattail victories did not represent a major shock to the American political system. As such, although the 103rd Congress and the presidency were controlled by the same political party, the government was in a period of continuous policy gridlock, with abandoned policy proposals and incremental changes. It was a unified government with unified gridlock. just as continuous policy inaction during a public outcry for change can lead to electoral turnover, so too can there be electoral repercussions from forcing policy action. As will be shown in later chapters, the party‑based pressure on some conservative Democrats to support Clinton proposals (including the 1993 budget act) led many members to vote against their constituents' preferences; this, in turn, led their constituents to vote against them in 1994, bringing about significant Republican gains.

These few examples of electoral reactions to policy change (or lack of change) show the dynamic nature of elections and policymaking. Exogenous shocks to policies or preferences may lead to inaction or to new policy outcomes. Such legislative action (or inaction) has electoral implications. And these elections in turn act as even further shocks to the system. It is a constant struggle for politicians to please the electorate while attempting to make compromises to break gridlock. We are describing revolving gridlock. Although legislators' preferences and institutional constraints always lead to a certain amount of gridlock, politicians and the electorate are always attempting new and different tactics to bring about what they perceive as good policy. Politicians are either caught in gridlock or making dramatic compromises in the attempt to overcome it. The public, when dissatisfied with congressional and presidential activities, can do little but vote them out of office.36 Unified government was thus attempted in 1992, and Republican control of Congress in 1994. And yet, due to the reasons discussed here, this revolving of politicians did not bring an end to policy gridlock.

The Role of the President

Members of the Bush White House were pleased with their cleverness. Democratic leaders had held off sending the minimum wage bill to the White House. Congress cleared HR2 on May 17, 1989, but the Democrats wanted their new House leadership in place before engaging in a confrontation with the President. On June 13, that leadership was in place, George Bush was out west on a two‑day trip, and the Democrats felt that it was time. They sent the bill to the White House, and scheduled a press conference for 3 P.m. There, they would make arguments intended to give the President great discomfort if he actually vetoed the bill, which raised the minimum wage above his $4.25 target.

But Bush was ready for them. His staff had made certain that he carried his veto message with him to Wyoming. At 1:39 P.m. the President was informed aboard Air Force One that the bill had reached the White House. He immediately vetoed it, having members of his team in Washington finish the paperwork to send the bill back to Capitol Hill half an hour before the Democrats' press conference. "This may be the first faxed veto in history," White House chief of staff John Sununu joked.

The President was confident about his veto, knowing that neither the House nor the Senate had passed the bill by the two‑thirds needed to override it. Additionally, he had a promise from more than thirty‑four Senators that they would sustain his veto. Now if only the American people would realize that he was committed to a wage increase, but not the bloated one that had passed through the Congress ... By undermining the efforts of the Democratic leadership to get the first punch in, Bush had taken a large step toward his objectives.

In previous sections we explain how the theory of revolving gridlock is based on the preferences of members of Congress and on congressional institutions, without much mention of the role of the President. Indeed, we feel that the President's role in policy making is less significant than he is often given credit (or blame) for in the press and in the minds of the voters. Nevertheless, the President does play an important role in the legislative process, a role that can be brought to light through the revolving gridlock theory. He affects lawmaking in at least four areas: influencing the preferences of legislators, vetoing legislation, compromising with pivotal legislators, and raising the importance of issues he would like Congress to act upon.

The President and his party can influence the preferences of members of Congress. A popular President can help members of his party through fund‑raisers, media publicity, and campaign activism. In close votes, the President and party leadership can offer side benefits to legislators in exchange for their votes, when they are indifferent to the issue at hand.37 These benefits could come in the form of pork or of campaign financing and publicity. Additionally, a popular President can provide cover for members of Congress who are nervous about policy votes. By taking a public stand, the President provides the opportunity for legislators to say that they are voting with the President on the issue at hand rather than having to explain their position in greater detail and thus upsetting a segment of their constituency.

The most formal involvement of the President in the legislative process comes through the use of the presidential veto. As described and illustrated above, the possibility of a veto stretches the gridlock region in the direction of the President's position. This means that policies dose to the President's position often cannot be made more moderate even if a majority in Congress would prefer such a change. Furthermore, with a change in the party of the President, the direction of the gridlock region's expansion changes, often releasing policy areas that had been left alone because of veto threats by a President of the other party. in addressing these newly released policy areas, the President has a chance to gain popularity and appear active. When a newly elected President replaces a President of his own party, the gridlock region does not change (the veto pivot is still on the same side of the median). As such, not as many policies are found outside the gridlock region for the new President to act upon. This situation undoubtedly added to the impression that Bush, who replaced a President of his own party, was ineffective in the legislative arena.

When the minimum wage was raised in 1977, the constraint on this increase was the possibility of a filibuster in the Senate. Had a conservative President been in office, the constraint on policy change would have been the more‑restrictive veto threat. Had Gerald Ford been reelected in 1976, the minimum wage hike would have been smaller than the $1.05 increase that President Carter secured. Likewise, had Michael Dukakis been elected in 1988, the ninety‑cent increase allowed by President Bush would have been surpassed. As the theory suggests, institutional conditions and the preferences of members of Congress always constrain the size and substance of a policy change. The greater these constraints (veto threats rather than filibuster threats), the greater the cries of gridlock. In minimum wage policy, we may see a difference of a few dimes in wage increases because of the position of the President. In dealing with the defense budget and entitlements, the difference may be billions of dollars.

The third role of the President in the revolving gridlock theory relates to policy compromises made with the pivotal legislators in the model. Recall from Figure 2.4 that the policy shift from the status quo to a point inside the gridlock region could result in raising the minimum wage to any amount between $4.00 and $4.65 an hour. In this case, $4.00 was the ideal policy point for the veto pivot and $4.65 was the position that made the veto pivot indifferent to selecting between the old policy and the new. Depending on the compromise made with the veto‑sustaining minority, any new minimum wage between $4.00 and $4.65 could be passed. By appearing ready to veto wage increases beyond his request of $4.25 an hour, President Bush could pull the outcome closer to $4.00, making a better deal for hiMSelf.38 But if the President had taken the unpopular position of opposing any increase, he may not have been able to gain the votes necessary to sustain a veto. Indeed, he may have been forced to give in to a policy outcome nearer the opposing legislators' position than his own. This would have led to the appearance of caving in and thus to a drop in his popularity. Even when a President is able to influence the policy outcome, however, he seldom gains as extreme an outcome as he might want. Based on the model, Bush was faced with a policy outcome of between $4.65 and $4.00, whereas his ideal minimum wage was probably less than $4.00 an hour. A President is always forced to accept policy results within the narrow band of the gridlock region. By taking a position of $4.25 and vetoing any increase beyond that point, President Bush won the bargaining game with the congressional leadership. In the end, the Democratic leadership was more interested in passing the increase that Bush proposed than in risking a second loss to the President on another sustained veto.

The final significant role of the President in the theory of legislative policymaking is in bringing the public's attention to an issue. Congress often has a number of issues on its plate at the same time. When a President emphasizes an issue during a campaign, a press conference, or the State of the Union Address, he is looking to bolster public awareness and encourage legislative action. Where there is a large degree of uncertainty surrounding a policy area, Congress may be hesitant to act without a push from the President. It is possible that the health care debate raised in 1994 would not have occurred without all of the publicity generated by Bill and Hillary Clinton. Whereas the revolving gridlock theory argues that legislator preferences and institutions will affect where policy will end up, the President can help decide which issues will be addressed. Thus when claims were made that the Whitewater accusations would affect health care policy, the reality is that, if Whitewater had any impact, the impact was on whether health care was addressed at all rather than how liberal or conservative the health care bill was in the end. From this view, the President becomes more an agenda setter than a force influencing policy outcomes. This is not to say that once Congress starts dealing with an issue the President can back off. If he does, the bill could easily die before it ever reaches the floor. Rather, the President must keep the issue in the public eye and help to resolve uncertainty about the issue through task forces and the advice of experts and executive agencies.

Opposing Theories

To further clarify the theory we are setting forth, we have found it useful to contrast the revolving gridlock theory with two others that have been circulating in media and academia in the 1990s. These theories have been used as assumptions in many works, but have not been fully addressed in any one work. The first is the "strong party theory," in which the control of Congress and the presidency by one party is expected to result in policy outcomes reflecting that party's ideal platform, whereas control of these two branches of government by opposing parties leads to gridlock. The second is the "compromise between branches theory," in which the Congress and the President strike a compromise between their ideal policy outcomes. According to this theory, a liberal Congress and a conservative President might agree on moderate policies, whereas a liberal President and a liberal Congress will certainly come to a (liberal) political compromise.

The assumptions of the strong party theory underlie many arguments about divided and unified government. Either explicitly or implicitly, those who write about divided government believe that control of the executive and legislative branches by different parties has policy consequenceS.40 Political scientists (Fiorina 1996; Jacobson 1990; Cox and Kernell 1991) have tended to be explicit on the causes of divided government and to be less explicit about the consequences. Broadly speaking, however, the consequences can be sorted into two kinds of claims: (1) divided government makes an already unwieldy constitutional system of government unworkable; and (2) divided government obscures responsibility.

Those who focus on the claim that divided government makes the system of governance unworkable are essentially arguing that divided government yields policy gridlock. Sundquist compares divided government unfavorably with a unified model of government that assumes an active President supported by cohesive legislative majorities. Essentially, divided government gives each branch of government incentives to undermine the actions of the other branch (Sundquist 1988, 629‑630). Ginsburg and Shefter (1990) make a similar argument, claiming that in divided government, governing becomes posturing and decisions satisfy no one. Although scholars who fit into the gridlock mold tend to be general rather than specific about policy consequences, some have claimed specific effects. Lloyd Cutler, for example, claims: "In modern times high deficits have occurred only with divided government.... The correlation between unacceptably high deficits and divided government is much too exact to be a coincidence" (Cutler 1989, 391). McCubbins (1991, 83‑111) corroborates this analysis, arguing that the high deficits between 1981 and 1987 were the result of party preferences and divided control of government. Alt and Lowrey (1994) look for a connection between divided government and budget deficits in the states. McKenzie and Thornton (1996, 157) agree with the general thrust of these arguments, claiming that divided government exacerbates the problems of dealing with deficits.

Those who claim that divided government diminishes electoral accountability argue that under divided control citizens cannot tell who is to blame; thus the meaning of electoral outcomes is rendered confusing (Fiorina 1996). This line of reasoning is so enmeshed in the normative argument about responsiveness and responsibility that sorting it out depends upon one's values. Thus we deal here only with the gridlock argument.

The major case against those who believe that divided government has policy consequences is made by David Mayhew in Divided We Govern (1991). In this work Mayhew argues that there is no relation between divided government and significant policy results; he further argues that there is no relation between divided government and congressional investigations of the executive branch. Mayhew concludes by arguing among other things that divided government does not mean less coherence in individual laws, and that there is no evidence that those initially less well off are made worse off by divided government (Mayhew 1991, chap. 7). Mayhew's findings contradict those scholars who assume that, without a strong President and cohesive congressional majorities, policy gridlock ensues. Mayhew, however, never describes the underlying model of government that would allow the passage of significant legislation even under Republican Presidents and Democratic legislatures.

Looking at the strong party theory in conjunction with the institutional constraints of the filibuster and veto in the revolving gridlock model yields a surprising result. Instead of bringing about an end to gridlock, strong parties actually expand the region in which gridlock occurs.42 Imagine the Democratic and Republican Parties deciding what policy outcomes they would like. One would be rather liberal, the other rather conservative. If the status quo policy falls somewhere in between, a movement to the right would be halted by a Democratic filibuster (or a sustained veto from a Democratic President). A movement to the left would likewise be successfully filibustered by the Republicans. With each party having more than forty seats in the Senate, the strong party assumption not only expands the gridlock region, but it also predicts the same region regardless of the party of the President. The gridlock region for strong parties stretches from one party's ideal policy point to the other's. Thus if we observe gridlock under Bush, we should expect the same gridlock under Clinton.

In 1992, the media and the Clinton campaign's rhetoric led the public to believe that Democratic control of the Congress and the presidency would end the policy gridlock of twelve years of Republican rule. There are two main problems with such an argument. The first is that this strong party model ignores the institutional constraints of the filibuster and the veto. When these constraints are included, the gridlock region actually expands out to the party ideal policies (and not just to the ideal policies of party moderates), leading to the likelihood of even more gridlock than under the weak party assumption that we have used. (Such an argument led many Democrats to the further belief that they needed sixty Democrats in the Senate to overcome Republican filibusters.) But the second problem with this theory is more dramatic: political parties in the United States are not necessarily strong. Members often defect from their parties when it is in their interest to do so. Republican filibusters are broken when the policy under debate is made conservative enough to make vote switching favorable for the most liberal Republicans. Indeed, in the minimum wage example, Senator Chafee abandoned the Republican filibuster attempt in 1988. Presidents are often forced to look for votes from the other party, as Reagan did with southern, Democrats, as Bush did with the 1990 budget deal, and as Clinton did with the North American Free Trade Agreement (NAFTA). When the House sustained President Bush’s veto of the minimum wage increase, the President lost twenty of the most liberal Republicans, but gained twenty‑eight Democrats, twenty‑five of them southerners. It is our contention that the appearance of party strength comes in the alignment of preferences: conservatives are typically Republicans and liberals are typically Democrats. Only where party interests are in conflict with personal preferences would we expect to see a break with the party; such breaks, however, are commonplace. The revolving gridlock theory relies on personal preferences and constraining institutions to explain policy gridlock and the positioning of bills that do pass. The strong party theory is much less explicit about the role of institutions, and has been brought into question by Mayhew and others.

Having discussed the strong party theory, we turn to a comparison of the revolving gridlock theory with the compromise between branches theory. This theory is also implicit in many works, but explicit in few; it supposes that policymaking power rests with both the President and Congress. The policy outcome will therefore be somewhere between the ideal policy of the President and the ideal policy of Congress; where in this region it will fall depends on the relative strengths of Congress and the President during the period in which policy is being formulated. If a President is strong and popular, the theory argues, he can publicize his ideal policy and gain concessions from Congress. If weak, he will be forced to the sidelines, as Congress works through the details. This theory of compromise between the branches of government is applied regardless of the party affiliations of the President and the majority in Congress, thus differentiating it from the strong party theory.

The theory's limitations are to be found in the question of where this presidential power arises in the legislative arena. A number of claims have been put forth in the literature. Some scholars claim that presidential power is the power to persuade (Neustadt 1960). According to this argument, the President can try to lead members of Congress to his point of view. When legislators are reluctant to come around to his line of thinking, the President can "go public" on the issue, hoping that constituents will persuade legislators that it is in their best interest to go along with the President (Kernell 1993). Other authors argue that presidential power originates in veto power (Rohde 199 1). In essence, a President can claim that he will veto any bill that does not meet his requirements, as Clinton did with universal coverage for health care. But this argument ignores credibility concerns; if Clinton had received a health care bill covering most Americans, would he have vetoed it, settling for no bill over a bill that came dose to what he wanted? Still others synthesize these presidential powers. Jones (1994) emphasizes a President's strategic position as it relates to public popularity, electoral mandates, and the lawmaking sequence.

The revolving gridlock theory gives the President the power to influence legislators' preferences by persuasion or by going public. It also gives the President the crucial veto power. But it does not assume that he will act in any way other than in his own interest. When a bill that has passed through Congress is closer to the President's preference than the status quo policy, he will generally sign it; when it is further, he will veto it. The result of the revolving gridlock theory presented in this chapter is not simply a compromise between Congress and the President, but a constrained policy within the Congress, with the President doing what he can to influence policy around the edges. Often, due to institutional constraints, a policy outcome is more conservative (or more liberal) than both the median member of Congress and the President would prefer; but it is the best outcome they can get, so they agree to it.

The revolving gridlock theory is a significant advancement over these prior theories, explaining executive‑legislative relations in a system of relatively weak parties in a way that is unique and compelling. The theory assumes that legislators in the House and Senate will attempt to move policy toward their preferred outcomes. They are constrained in doing so by supermajority institutions. Particular members of Congress in key positions at the edges of the gridlock region enforce these constraints through support of filibusters and vetoes. Status quo policies in this region cannot be changed. Actions taken under uncertain conditions as well as exogenous; shocks to the legislative system can knock policies out of the gridlock region. Congress will be able to bring these policies back inside the region only so far as is possible to keep the pivotal members from voting down cloture or sustaining a veto. Attempts to make more extreme policy shifts will be killed by filibusters and vetoes. This theory is an improvement on previous work in the field. It looks at legislators' preferences rather than simply at their partisan affiliations. It takes into account the constraining institutions ignored by the strong party theory. And it more clearly defines the nature of the compromise between members of Congress and the President.

The theory explains minimum wage policy formation quite capably. Throughout the 1980s, the minimum wage was losing purchasing power, causing the $3.35 level to fall out of line with the preferences of most members of Congress. This is an example of an exogenous shock (here, inflation) pushing a status quo policy outside the gridlock region. When the issue was raised in 1988, it was first used more as a political tool than as a target for policy change. The Democrats used policy proposals to call Bush's "bluff " on minimum wage policy, whereas the Republicans used the issue as an attempt to secure federal judicial positions. No bill was passed in 1988 because the situation was uncertain so close to the election. When would the economic downside to a wage hike come into effect? Would interest groups play a large role in the elections as a result of this issue? Would one side or the other be in a better position in the next session due to the election results? Congress abandoned the issue. In 1989 the federal government addressed the minimum wage once again, this time with less uncertainty. Bush proposed a wage of $4.25 an hour, and the Democratic leadership countered with $4.55. Either wage level was preferred by the vast majority to the current $3.35 wage. But the President was able to use the veto and his influence to win a victory for his minimum wage plan. As such, he appeared strong in his veto power and secured nearly the lowest wage politically possible. The revolving gridlock theory is clear as to why this policy outcome occurred. The constraint was the veto pivot to which Bush appealed with his veto of the higher wage. By issuing an alternative policy favorable to the veto‑sustaining minority, the President was able to win his showdown with the Congress, securing the $4.25 wage in late 1989. In 1996, President Clinton took his opportunity to raise the minimum wage further. The constraint on his wage hike was the filibuster in the Senate instead of the more constraining veto pivot of conservative Presidents.

Journalists and academics have recently taken greater notice of the policy implications of the supermajority institutions of the filibuster and the veto. For example, in the conclusion of a recent article about partisanship in Congress, Cooper and Young (1997, 269) note:

In the House a veto requires not merely 218 votes to overcome (if all members vote) but 29 1, and in the Senate 67 votes are required (if all members vote) not merely 5 1. These are very difficult levels to obtain in partisan Congresses unless majority margins are extremely high. Moreover, in the Senate, practice with respect to the filibuster has changed so that the 60 votes required in impose cloture are also required to win any major policy battle. As a result, the passage of major legislation still requires forms of behavior and negotiation that are coalitional, but in a context in which the character of party divisions provides poor incentives for such behavior. The public's disgust with paralysis may spur action when elections approach, but such a response is only the flip side of the political maneuvering to gain electoral advantage that dominates policymaking and usually stymies action.

The revolving gridlock theory expands upon and clarifies this argument, noting specifics of how these supermajority institutions, along with elections, budget constraints, and members' policy preferences, lead to specific policy outcomes or gridlock in the last two decades.

In the next chapters we will take a broader policy view, attempting to explain many of the legislative outcomes of the periods of divided and unified government from Reagan's first term through the 1994 capture of Congress by the Republicans, using the revolving gridlock theory. In Chapter 3, we will take an extensive look at the budget process and how budget policies have changed since 1980. The conservative members of Congress who came in with Reagan in 1981 acted as an exogenous shock, allowing previously entrenched policies to be changed. Even still, the policies enacted were determined by those near the middle of the political spectrum. As such, Reagan needed to gain the support of conservative Democrats. He used this support to enact a huge tax break and to peg tax rates to inflation. Budgetary policy since 1981 has been limited to small changes, as no new coalitions had been formed by electoral shifts until the 1994 election. The specter of deficits as far as the eye could see left politicians no room to maneuver to break the gridlock that ensued.

As we will highlight in the next chapter, the dramatic budget deficits of the 1980s have combined with supermajoritarian institutions to further policy gridlock in four ways. First, as mentioned above, pivotal members of Congress must vote with the majority to secure policy change. These votes are typically gained, either through adjusting the policy at hand toward these legislators' preferred outcomes or by guaranteeing support for these members on other bills, typically by providing them with pork in budget bills. With the budget wells running dry with the significant deficits of recent years, however, those seeking policy change found the latter option for gaining the pivotal members' support to be no longer available. As such, compromises that could be forged only through budgetary concessions now fell apart. Second, deficits led Congress to link together most budgetary decisions. Increases in one program area would need to be offset by decreases elsewhere or by increases in taxes. As can be imagined, this had a further impact in terms of policy gridlock. Once a supermajority coalition was established in support of a legislative proposal with budgetary implications, bill proponents had to convince coalition members to support tax increases or program cuts elsewhere. This was a significant enough constraint to lead to inaction on issues in which a majority or supermajority agreed that change was necessary. Third, the increasing difficulty and importance of budgetary decisions gave members of Congress even less time to deal with non-budgetary issues. Without the necessary time to secure supermajorities or to resolve uncertainties, other issues fell into gridlock as well. Finally, the complexity of budget legislation in an era of omnibus budget bills and huge reconciliation packages led to further confusion, frustration, and gridlock.

Clearly the budget constraints on Congress combined with electoral and institutional constraints to bring about further policy gridlock. It is to these budgetary issues that we turn next.

NOTES

1. The comments, quotations, and details presented in the italicized sections throughout this chapter are drawn from various public sources.

2. While our work emphasizes the preferences of members of Congress and the President, it does not emphasize the intensity of these preferences. This is largely due to our focus on voting institutions within Congress that give everyone equal say. However, as Hall (1996) points out, members of Congress feel more strongly about some issues than others, leading them to different degrees of participation in committees, subcommittees, and in the brokering of deals over final legislative decisions. As we note below, the intensity of a legislator's preferences may play a role in whether others may try to influence that legislator's vote, as well as in what final bargain may be reached among the House, Senate, and the President within the limited range of possible outcomes.

3. In this work we use the term "revolving gridlock theory," but we also find Krehbiel's tag, "pivotal politics theory," accurate and appropriate.

4. Political scientists have used spatial models widely since Anthony Downs's An Economic Theory of Democracy (1957) and Duncan Black's The Theory of Committees and Elections (1958). Single‑dimensional voting with a majority rule was found to lead to median voter outcomes. However, without such a limitation to a single dimension, even imposing a particular voting rule often led to indeterminate outcomes (see Arrow 195 1; Black and Newing 195 1; Plott 1967). The chaos result formalized by McKelvey (1976) indicated that in all but a few special cases a series of proposals could be developed to lead from any given policy status quo to any other policy in the choice space. This result troubled formal theorists more than it troubled most political scientists who, without empirical support for this "policy cycling" result, discounted the value of such spatial models. The response to this "anything can happen" view was the reassurance that there are political structures in place that keep such ludicrous results from happening. Kenneth Shepsle's "Institutional Arrangements and Equilibrium in Multidimensional Voting Models" (1979) led to an argument about which structures or institutions are relevant in leading to various outcomes. In particular, do legislative committees with their proposal powers lead to agendas that provide them with beneficial outcomes? This article set off a debate on the power of committees, as well as on which institutions lead to the so‑called "structure‑induced equilibria." Major groundwork for further study of politics through spatial models was made by Romer and Rosenthal (1978) with regard to the control of legislative agendas, by Baron and Ferejohn (1989) with regard to the sequencing of proposals leading to political bargains, and by Gilligan and Krehbiel (1987, 199o) with regard to the role of information in legislative decisions. For an excellent review of these advancements through the 1980s, see Krehbiel (1988). Our view is that in the modification of spatial models in political science over the past few decades, some simple advancements and applications have been overlooked. In particular, even restricting the model to a single dimension, much can be gained by looking at the institutional structures that require supermajorities to pass legislation. See Krehbiel (1997, chaps. 7, 8) for an excellent analysis of how the single‑dimensional pivotal politics theory works in the face of empirical findings regarding agenda setting, partisanship, uncertainty, and presidential influence.

5. Poole and Rosenthal (1991a, 1997) provide an excellent compilation of voting in Congress and thus legislator preferences.

6. Binder and Smith (1996) provide the most comprehensive analysis of filibusters to date. For a further summary of the use of filibusters throughout the history of the Senate, see the 1987 Congressional Quarterly Weekly Report, 2115‑2120. For more information on the rise of the filibuster and other changes leading to a more individualistic and competitive Senate, see Davidson 1985 and Sinclair 1989.

7. Krehbiel and Rivers (1988) use such an alignment in their analysis of committee positions and proposals with regard to minimum wage.

8. For a similar argument of how the press portrayed the economy in 1992 and how it led to BusEs defeat, see Hetherington 1996.

9. Some theories would suggest that members of both houses err on the side of willingness to use blocking tactics in order to avoid blame, on the assumption that avoiding blame is more important than taking credit. See, for example, Weaver 1986 and Arnold 1990. We raise some of these concerns below in our discussion of uncertainty.

10. What is important to us here is the near indifference of this pivotal legislator to the proposed policy change. This indifference may make this pivotal individual the target of persuasion to either cement the deal or sabotage it. Although we do not explore the matter of legislator preference intensity here, in a similar fashion, legislators who have a low intensity of preferences (for whom the issue at hand is not very salient) might also be easilY Persuaded to vote for or against the legislation. For more on legislator preference intensity and the resultant participation decisions, see Hall (1996).

11. Note that policies in the region between Q and F are outside the gridlock region, and thus will not be adopted by the Senate.

12. This result is different from that derived by Krehbiel (1996). Krehbiel limits the interaction to a single veto and override attempt. In actuality, legislation that is vetoed may be attempted again in a revised form. We do not derive results from a formal game of this riature here, but simply note that the ensuing action is a form of bargaining in which the resultant outcome is in the range between the main actors' ideal points and is dependent on their bargaining strengths. Additionally, if the President or congressional committee makes a proposal in this range under a closed rule, it will be accepted by the pivotal members who prefer the proposal to the status quo; while under an open rule, the median member and concurring majority will push policy toward the median (although the policy is still constrained by Q'). As such, our finding is not inconsistent with the seminal agenda‑setting work of Romer and Rosenthal (1978).

13. The intensity of preferences may play a role in influencing the patience of the actors who participate in the bargaining (Hall 1996) and what final outcome is reached.

14. The minimum wage increases ftorn left to right on the diagram. Therefore the figure does not fit the traditional left‑to‑right spectrum, in which conservatives are positioned on the right and liberals on the left. Here, the position of Bush and the conservatives is on the left, favoring a lower minimum wage. We hope that this arrangement is less confusing than the inverse, with dollar amounts decreasing from left to right.

15. This assumption is justifiable in hindsight, based on voting behavior, but may have been a bit trickier to judge at the time. Nevertheless, Democratic leaders had proposed a $4.65 minimum wage, claiming that they could collect enough votes to override a veto. This would be the case with a veto pivot at $4.00, where that legislator would be indifferent to whether the minimum wage were 65 cents higher or 65 cents lower. Volden (1997) explores the positions of legislators (and the possibility of their voting strategically) with regard to the 1989 minimum wage vote, corroborating the values set forth in this chapter as quite accurate in representing legislator preferences.

16. Gilmour (1995) notes conditions under which Congress might provoke vetoes or otherwise promote stalemate through strategic actions.

17. On legislation that cannot be filibustered in the Senate, the gridlock region in the Senate becomes like that in the House, stretching from the median to the veto pivot. Such legislation includes budget reconciliation as well as trade bills set on the "fast track." This smaller gridlock region acts as less of a constraint to policy change, thus allowing for more policy action and limiting the power o a minority to stop egis action. While the lack of a filibuster threat on such legislation affects the size and shape of the gridlock region, it does not affect the overall theory with regard to this region. Status quo policies in the gridlock, region cannot be changed. Those outside of the region will be brought in, with limitations similar to those seen in Figure 2.3. The theory surrounding this smaller gridlock region will be clarified further in the next chapter, which attempts to explain the major budget' changes that have passed Congress since 1980.

18. Note that Krehbiel's work (1996) does not contain uncertainty. Nor do we introduce," uncertainty in a formal modeling sense here. Our point is that uncertainty and the complexity of forming legislation may lead to further policy gridlock under certain conditions. This thesis is explored in greater detail in later chapters.

19. Where this uncertainty cannot be suitably resolved, legislators might try blame_~' avoidance tactics. See Weaver 1986 and 1988, and Arnold 1990.

20. For an analysis of the information provided by committees and the surrounding incentives to gaining expertise, see Krehbiel 1991.

2 1. Reported in the New York Times, July 6, 1988.

22. For more on the pressures surrounding this decision, see the 1988 Congressional Quarterly Almanac, 255.

23. Reported in the New York Times, December 8, 1978.

24. These decisions are part of the strategic choice process followed by politicians with regard to elections. See Jacobson and Kernell 1982; Jacobson 1983 and 1989; and, Rosenstone and Hansen 1993.

25. As reported in the New York Times, September 19, 1988, members of Congress were being very careful about their positions on minimum wage.

26. Stopping policy change does not necessarily mean that no bill will be passed. For political reasons, legislators could pass a "hollow" bill, one that does not change policy to any great degree.

27. Thus, in focusing only on expansions and contractions in the gridlock region, Krehbiel (1997) is missing part of the story, that of electoral shifts.

28. See Brady 1988.

29. See Erikson 1976; Fiorina 1991b; Jacobson 1991; Ansolabehere, Brady, and Fiorina 1992; and Alford and Brady 1993.

30. Our view, is not inconsistent with that held by Alesina and Rosenthal (1989, 1995) who argue that voters at the margins will choose liberal Democrats to counter conservative Republicans. Fiorina (1991a, 1996) presents the case that voters may want to elect politicians who will provide a "check" on other politicians. As a baseline against which voters might decide to temper policy outcomes, voters often perceive particular politicians as espousing particular issues or viewpoints. Jacobson (1990) and Petrocik (1991) present a form of "issue ownership" argument. Jacobson argues for the public perception that Democrats might better address local problems and Republicans national issues (although with Clinton as President, the opposite might now hold true). Petrocik argues that each of the parties owns a set of issues, with Republicans seen as preserving low taxes and pursuing prosperity and Democrats seen as espousing kindness through social spending.

31. The "coattails" effect refers to the election of other members of the President's party to their respective offices due to a surge of presidential supporters going to the polls. See Ferejohn and Calvert 1984.

32. See Erikson 1988.

33. See Brody 1991 for views on the role of the "honeymoon" in assessing a President.

34. Rivers and Rose (1985) discuss how the size of a President's program affects its likelihood of success.

35. Weaver (1,988) attempts to explain conditions under which Congress indexes programs to inflation, leading to "automatic government" without the need to continually deal with the exogenous shocks of inflation.

36. Hibbing and Theiss‑Morse (1995) explore this public dissatisfaction with Congress, finding that the actions and activities of politicians disgust the public equally regardless of the political party. Politicians of both parties are seen as being identical in the types of actions they take, regardless of the distinctions they attempt to make on particular policy issues. This is consistent with our belief that politicians act on their own preferences, using the institutions of government to their best advantage. Often this is to the best advantage of their constituents as well, and, when this is perceived by the public, the politicians are reelected.

37. The effectiveness of using presidential popularity to aid members of Congress is brought into question by Collier and Sullivan (1995).

38. Volden (1997) argues that this bargaining process is complex, often leading to members of Congress voting against their immediate preferences in what is referred to as a "sophisticated vote, '

39., Important theoretical and empirical work on this issue has been advanced by Samuel Kernell (1993).

40. See Fiorina 1996 for an excellent summary of the literature pertaining to divided government.

4 1. James Pfiffher (in Thurber 199 1) argues that unified governance with strong parties is necessary for directional coherence of legislation.

42. For further details and a formal proof of this finding, see Krehbiel 1996.

43. Some views of the strong party assumption consider only a strong Democratic party. This would still lead to an expansion of the gridlock region when compared to our weak party assumption of individual preferences mattering more than the party preferences, but the expansion would be only to the left.

44. It is not our intention here to become enmeshed in the debate about how strong or weak political parties are. It is clear to us that, as Cox and McCubbins (1993) suggest, parties can use various powers, acting as "legislative cartels," to gain mutually beneficial compromises. We also recognize, as Krehbiel (1993) does, that parties are limited in their abilities and that a strict test of party strength is a difficult endeavor. Our view of political parties in the United States today is that party leaders and the President can influence the preferences of members of Congress, but cannot dictate how they will vote. Party members also have other interests, mainly reelection. We believe that parties are weaker than they once were (for a view on parties at a time when they were stronger, see Brady and Epstein 1997). We also see a strong link between preferences and party~ with conservatives tending to be Republicans and liberals tending to be Democrats. (In Krehbiel's view this is exactly what parties are‑aggregated preferences.) In some ways our view is not unlike that found in the political science literature of the 1960s and 1970s. Following E. E. Schattschneider's responsible party thesis (1942, APSA 1950), David Truman's work (1959) reversed the strong party notion by focusing on congressional parties as blocs of voters with party leaders near the party's center. We speak of preferences where Truman, as well as Burns (1963), focused on blocs or wings, but their view of parties as aggregates of different views is not unlike our own.

45. In this way, our work complements that of John Gilmour (1995), who argues that the need to satisfy constituents (preferences) often leads to pursuit and avoidance of ideas and proposals, with politicians provoking vetoes (supermajority institutions) and taking positions that cause greater difficulty in the negotiations necessary to reach compromise (gridlock).