AMERICA HAS A fundamental problem:

Because power in our political system is so diffused, our national government is becoming incoherent and irresponsible. It has become virtually impossible to sustain national leadership; we get new administrations, but they cannot build congressinal majorities that will legislate their policies. The problem is now widely acknowledged, but most of the proposed solutions involve sweeping and difficult constitutional change.

However, there is one idea that does not involve a constitutional amendment, is relatively easy to bring about, and might be dramatically effective. We could reform the rules of each party's presidential nominating convention so that so that the party's congressional wing would play a key role in selecting the party's presidential candidate. With such a change, the winning party would have a much better chance of forming a coherent national government that might actually carry out a coordinated program.

The objective would be something closer to parliamentary government than is now possible -- something that, without altering our constitution, would give the United States the kind of government it deserves to cope with the complexities of our age.

The reform itself is simple: creation of a bicameral nominating convention for each party. One chamber would consist of delegates chosen by caucus and primary, just like the present-day convention. The new, second chamber would consist of all the party's chosen candidates for the House and Senate, plus the holdover senators.

To win the presidential nomination, a candidate would need to win a majority of each chamber. Should different candidates win in each chamber, there would be a further run- off ballot between them.

One of the oddities of our present political party system is the disconnection between each party's congressional wing and the process for nominating the party's candidate for president. Members of congress and candidates for Congress play little or no part in choosing the presidential nominee. This is one reason why the ties between incumbent presidents and incumbent legislators of the same party are normally so weak. Except in times of great crisis like depressions and world wars, the president and the legislators of the same party usually find it difficult to "form a government" that can adopt a coherent set of policies and stand accountable for the results.

The party presidential nominating machinery has become a passive instrument. Regardless of party experience and the views of party leaders, anyone with enough money, doggedness and political skill can use the existing machinery to gain the nomination before the convention begins. He can do so by winning primaries that usually turn not on which candidate best reflects the policies of the party or shows the best capacity to govern, but on which can buy more television time and present a more pleasing television appearance.

Congressional incumbents and congressional candidates play virtually no part in this process. As often as not in recent times, the presidential nomination goes to someone with no prior connection whatever to the congressional branch. Indeed, a congressional leader who decides to seek his party's presidential nomination faces two major handicaps -- lack of time to campaign and the need to cast votes on controversial issues that offend important segments of the electorate. Since World War II, the major parties have had 18 chances to nominate a candidate for president; only three of those nominated -- Kennedy, Goldwater and McGovern -- were incumbent members of Congress when nominated, and two of them lost.

Is it important whether a new president and his party colleagues in Congress can "form a government"? The framers of the Constitution sought to separate the president from Congress, and the two houses of Congress from each other. They created this system of checks and balances because they were far more preoccupied with the dangers of tyranny than with the virtues of efficiency. Over two centuries, only a handful of our presidents have been able to "form a government" with Congress -- usually in times of extreme crisis -- yet the Republic still stands.

But our world is very different from the world of 1787, or 1887, or 1937. In those earlier times, our national government was not expected to assure steady economic growth while preventing inflation, and while providing a measure of social justice for all citizens. Though it has always been the government's duty to assure the national defense, this did not in earlier times mean the defense of peace and freedom throughout the world.

Today, neither economic nor security policy can be unilateral. Trying to manage just the national economy of the United States in 1983 would be like trying to manage the economy East of the Mississippi River in the 1930s -- there's no such thing as a "national" economy left to manage. Security policy, similarly, must involve allies around the globe.

In today's complex and interdependent world, the ability to form a government that can cooperate effectively with other nations is crucial. Most other nations, whether totalitarian or democratic, manage to do so. It is no exaggeration to say we are the only major nation whose head of government cannot commit the government he heads. Other nations have learned by painful experience that an agreement with the president of the United States is often merely the beginning of a further negotiation with the Congress. In those instances where any binding agreement can be achieved, foreign states have had to learn to save some concessions for this second round of bargaining.

There is another reason why it is important to improve our ability to form a government. The purpose of periodic elections is to hold public officials responsible for their policies while in office, and to turn out those who disappoint us in favor of those who persuade us they have better policies. But whose policy was it that resulted in the largest deficit, the deepest recession, the highest unemployment -- or, in the previous administration, the highest inflation -- since World War II? Whose policy left the national government -- four times in the last three years -- without any new set of appropriation measures when the preceding ones expired?

We cannot say it was the president's policy, because Congress either disregarded many of his proposals and enacted different legislation of its own, or failed to agree on any legislation at all. We cannot say it was the policy of Congress, or even of the majority party in either chamber. Despite the real improvements in the congressional budget process, Congress still makes policy on an ad hoc basis, one issue at a time, and builds a different cross- party coalition on each issue as it arises.

The sum of all these actions is not a "policy" for governing; it is a hodge-podge for which neither the elected president, nor either party in Congress, nor any elected legislator in either house will accept responsibility. In the apt idiom of a 1930 prizefight character, "It's every man for theirself." As a result, the public cannot fairly hold any elected official to account. Each of them is adept at shifting blame to the others.

Fairly or not, we voters usually hold the president accountable. We blame him because, at least since the example of Franklin D. Roosevelt, we expect him to lead the nation. When he cannot persuade Congress to follow his leadership, we regard that as a personal fault. While presidents of course vary in their personal leadership qualities, it is a rare president whose leadership is so exceptional that he can overcome the basic structural fault in our system over an entire presidential term.

That basic structural fault can be simply described. To govern with any sense of direction, the president and a majority of each house of Congress must agree on the balance they plan to strike among the competing and conflicting goals of government, and the mix of taxation, outlays and other measures that fits this balance.

In the parliamentary governments -- including all six of our principal Free World partners, Britain, Canada, France, West Germany, Italy and Japan -- this is done by "forming a government." A majority of the legislature -- including all the members of one or more parties -- agree to stick together to legislate and execute a program under the leadership of a premier or prime minister chosen by the legislative majority (or, in France, of a popularly elected president). If the majority breaks apart over a key part of its program like the budget, the government "falls," and either a new leader is chosen by a new legislative majority or a new general election is held.

Because the majority does not like the consequences of "falling," its member tend to support the entire program of its leaders even when they disapprove of particular elements. When the results do not please the legislature or the voting public, accountability is clear, and a change in both leaders and policy usually occurs.

In our system, the president and the legislators have little involvement in one another's election or reelection. Even when they are members of the same party, they have little incentive to agree on what the elements of a balanced program should be. The ultimate incentive -- sharing the same political fate if the program succeeds or fails -- is much weaker than in parliamentary governments, because legislators so often win reelection even though their party loses the presidential election.

In 1980, for example, 87 percent of the incumbent Democratic senators and congressmen who were renominated won reelection, even though President Carter was decisively defeated and lost a majority of the same states and districts that these Democratic legislators carried. In 1952, 1960, 1968, 1976 and 1980, the average percentage of renominated senators and congressmen reelected despite their party's loss of the White House was a whopping 92 percent. Each party's incumbent legislators manage to avoid sharing their presidential candidate's electoral fate.

So perhaps it is time for a party reform -- not to create an American version of a parliamentary system, but simply to improve the chances of giving America a coherent government. We can do this by giving the congressional wing of each party a major role in nominating the party's candidate for president, as they did in the days of Adams and Jefferson.

Last year a Democratic National Committee Commission on Presidential Nomination found that the party's elected officials "have not participated in and thus have felt a limited responsibility for our recent national conventions." It noted that out of over 300 Democratic members of the House and Senate, less than 15 percent participated as voting delegates in the 1980 convention.

As this commission recommended, the Democratic Party has begun to give the congressional element of the party a greater voice and make the convention itself more deliberative. Beginning in 1984, House and Senate Democrats will select 60 percent of their number (or about 7 percent of the 3,800-odd delegates) to be unpledged delegates at the convention.

This is an excessively modest step. The parties can go much farther. And under a 1980 Supreme Court decision (Democratic Party of U.S. v. Wisconsin ) the changed rules could not be overridden by federal or state election statutes; within constitutional limits, the parties enjoy total freedom of action in the nominating process.

Here's how the rules might be changed:

Like Congress, the presidential nominating convention could be divided intootwo chambers. One -- the popular chamber -- would consist of the delegates as presently selected, minus those entitled to seats in the second chamber.

The second -- the congressional chamber -- would consist of the party's candidates for election to the House of Representatives and the Senate, as selected in the state primaries and caucuses up to the date of the presidential nominating convention, plus the holdover senators.

To be nominated as the party's candidate for president and vice president, a nominee would need a majority vote of each chamber. The chambers would meet simultaneously but separately. Each chamber would deliberate and vote until one candidate wins the support of a majority. If the same candidate wins in both chambers, he or she would be the nominee. If each chamber picks a different candidate, each would then conduct a runoff ballot between the two winners.

A nominee would then be chosen if one candidate obtained a majority in both chambers' runoff ballots, or -- if the two houses could not agree -- by calculating which candidate received the highest percentage of votes in both chambers added together.

For example, if Smith won over Jones in one chamber by 60 percent to 40 percent but lost to Jones in the other chamber by 44 percent to 56 percent, Smith's average would be 52 percent, Jones's average would be 48 percent, and Smith would win the nomination.

The party's platform would also require the concurrence of both chambers, with differences between the two chambers resolved by a conference process akin to the one now used between the House and Senate.

A scheme like this would produce a genuinely deliberative convention; delegates would have to use judgment. The members of the congressional chamber would not be bound by a primary or caucus to any candidate. Delegates to the popular chamber would be bound to candidates only to the extent presently allowed by party rules or state law, but would have total discretion if a runoff ballot between the two chambers became necessary.

Under such a plan the candidates would have a better chance of forming a government than either party's candidates have today. A presidential nominee would need broad support from his party's congressional candidates to win the nomination. A presidential nominee who prevails in the general election should feel more responsive to the party's successful congressional candidates and the holdover senators, and they should feel more accountable to the voters for what the incoming administration does.

There are many possible objections to this idea. Some will say the people's voice, as expressed in the state primaries and caucuses, should be controlling. But the congressional chamber will consist primarily of candidates who have themselves been selected by popular choice, and they should be trusted with a share of the responsibility for selecting the party's candidate for president.

One could argue that a congressional chamber consisting of all of a party's nominees for the House and Senate would give too great a voice to hopeless candidates for the 100-odd seats that are usually safe for the other party, and that this chamber will not be truly representative of the party's "real" congressional wing. But the number of safe seats is dwindling, and it seems unlikely that the party's probable losers in the general election would vote much differently from the probable winners and holdover senators. (One exception might be Republican candidates for safe Democratic seats in the South, who would probably be more right-wing than national Republicans as a whole.)

Others may object that a bicameral convention would be procedurally too cumbersome and might last too long. But our present conventions have become a monumental bore; many of their processes could be condensed to advantage.

For example, most of the set speeches could be eliminated, and electronic voting or written ballots could replace the interminable roll-calls. (Indeed, this would be a necessary measure in any run-off ballot between different winners in each chamber, in order to prevent inflation of the winner's percentage margin by vote changes as the roll-call nears its end.)

Finally, there is room for doubt whether a stronger congressional role in nominating a party's presidential candidate and writing its platform would really assure better cooperation between the congressional and presidential candidates elected to office. The institutional rivalry between the two branches would still be formidable. There would be no guarantee of party cohesion within Congress, and incumbent legislators would still be tempted to desert the president and save themselves when public expectations are disappointed.

But the bicameral convention would at least do something to draw a president and the legislators of the same party more closely together, and make it more difficult for them to blame one another for failure. The nominating process would also become much more deliberative, and would be much more appealing to the public. It might produce better presidential candidates who know more about how to govern rather than how to win primaries. It might even encourage a party's presidential and congressional nominees to exchange mutual pledges that if elected they would cooperate to "form a government" and to campaign on the basis of that pledge. (The Republicans set a good example for this sort of pledge in 1980.)

We need to do something dramatic to alter the dreary stalemate of the last two decades, in which the sum of conflicting government policies produces results which neither the president nor any legislator wants, and for which none of them is prepared to take responsibility. Before things get so bad that we have to consider changes of a constitutional dimension, the parties ought to give the bicameral convention a try.