American elections have long best been understood as referenda on the economy, but last November the economy didn't seem to matter to voters. By most traditional measures 1992 through 1994 were years of strong economic performance, yet the incumbent Democrats took a historic beating. For exultant Republican pollsters, this was evidence of a genuine ideological victory. For social scientists who touted models of voter behavior geared to economic aggregates such as growth in gross domestic product, it was another professional embarassment. For Democratic strategists, or anyone interested in something better than Contract politics, it should be a wake-up call for a more careful analysis. Either the economy really does not matter, in which case Democrats need to find new ideological appeals to an increasingly conservative voting public. Or it matters in ways that traditional analyses do not detect, in which case Democrats need a new electoral strategy based on new economic appeals.
Which is it?
THE IMPORTANCE OF THE ECONOMY
Wisdom here begins with an analysis of which voters deserted the Democrats in 1994, by categories of voters. Compared to 1992, Democratic support declined 10 percentage points among high school dropouts, 11 points among high school graduates, and 12 points among those with some college. Support held perfectly steady among those with college degrees. The same basic desertion pattern applies if the basis of comparison is the 1990 mid-term election. The decline in Democratic support again comes mostly from among the noncollege educated: down 9 points for high school graduates and 11 points for those with some college. The chief difference is that the 1990-1994 comparison shows very little change among high school dropouts. Merging the results, Democratic support appears to be curvilinear. Democrats are doing better at either end of the education distribution while collapsing in the middle.
Unfortunately for the Democrats, the middle is where most of the voters are. Exit polls, heavily used by strategists, understate the Democrats' problem. These polls showed the Democrats maintaining their support among college graduates, which appeared to be the largest single education group in the electorate (43 percent of voters, according to the Voter News Service [VNS] exit poll). But exit polls overcount better-educated voters. Evidently, college-educated voters are more articulate and more willing to cooperate with exit polls, which accounts for their overrepresentation. The more reliable voting data available from the Census Bureau indicate that college-educated voters were only 29 percent of the electorate in 1994. This is actually a smaller share of the active electorate than voters with just a high school diploma (31 percent), and far smaller than the overall noncollege-educated share (71 percent).
Among whites, where the decline in Democratic support between 1992 and 1994 was concentrated (black Democratic support actually went up slightly), the shift away from the Democrats among noncollege-educated voters was even more pronounced, especially among men. Between 1992 and 1994, Democratic support declined 20 points (to 37 percent) among white men with a high school education and 15 points (to 31 percent) among white men with some college.
Contrary to the hopes invested in the "gender gap," noncollege-educated white women also deserted in droves. For both white women with a high school diploma and those with some college, Democratic support dropped 10 points. Thus, to ascribe the falloff in Democratic support to "angry white guys," as many commentators did, is to miss the point. Large numbers of noncollege-educated white men and women alike abandoned the party.
What is true of the electorate as a whole is also true of the 1992 Perot voters, a key swing group. Adjusted to reflect voting patterns in the Census data, their ranks are even more dominated by noncollege-educated voters. And they moved massively against the Democrats, down 17 points from 1992 to just 32 percent support. If repeated in 1996, this astonishingly low level of support will doom Clinton's re-election bid. Indeed, recent polls show 1992 Perot voters breaking two to one against Clinton in trial heats for 1996.
And what does this all have to do with the economy? Approximately everything. As indicated in table 1, it is precisely noncollege-educated Americans, in particular noncollege-educated men, who have suffered the largest wage declines over the last two decades. Furthermore, this miserable experience of wage (and income) decline continued apace in the first two years of the Clinton administration, providing little relief to voters who have seen their living standards erode over time. (See Lawrence Mishel, "Rising Tides, Sinking Wages," ) Finally, analysis of wage data cross-linked to the VNS data indicates that the Perot voters who voted Republican in 1994 were the ones under the most economic stress, with estimated post-1979 wage losses more than double those of Perot voters who voted Democratic.
In short, the economy matters as much or more than ever, even if how it matters is missed by traditional election forecasting models that stress overall growth. These models mistakenly presume a "rising-tide-lifts-all-boats" economy that no longer exists. In today's economy, some boats get swamped by the same tide that elevates a few. Therefore, we need to look below the surface to see what is really going on.
HOW THE ECONOMY AFFECTS SOCIAL ISSUES
To be sure, wage data are only crude measures of how the economy affects individual voters, and narrowly material concerns hardly exhaust voter motivation. Along with their income, voters are naturally concerned about the stability of economic arrangements, the amount of time they get with their families, the quality of schooling that their kids are getting, the safety of their streets, the stability of their neighborhoods, even the robustness of those civic values on which any society depends.
Indeed, a broad definition of "living standards" would encompass all these factors, since they all influence the material quality of everyday life. All of these factors, noneconomic and economic, can and do independently motivate voter choice, with the noneconomic sometimes outweighing the economic. Still, social life is materially conditioned, and the basic material of social standing is income. Job security would be less of a worry with higher economic growth and full employment. Relentless abuse and stress at the workplace would not be tolerated if they were not thought necessary to maintaining income. Public goods would be more abundant if the taxes to pay for them were available, and taxes would be better tolerated if incomes were rising. Families and community life would get more time if less time needed to be spent working.
Thus, however enlarged a conception of "living standards" one favors, family-supporting wages and income are central. And a central fact of life for the vast bulk of noncollege-educated voters is that, increasingly, American politics does not assure it. Do that long enough, and you have an angry electorate. Do it again, and they will throw you out, almost whoever "you" happen to be.
Our quarrel is not with those who wish an expanded understanding of living standards and thus a Democratic focus on both economic and noneconomic issues. Indeed, we believe that such a synthetic approach is suggested by the 1994 election results and would be a big improvement over both the narrowly economic focus of traditional liberals and the basically noneconomic focus of orthodox New Democrats. Our real quarrel is with those who argue that 1994 marked a newly ideological turn in American politics, from which the decline in living standards can presumably be divorced. This is a widespread view in the mass media, which take 1994 as a general right turn. It's true that the percentage of self-identified conservatives increased from 30 to 37 percent between 1992 and 1994. Republican pollster Fred Steeper points to the unusually high Republican House vote among self-identified conservatives (81 percent) in 1994--a 9-point increase over 1992--as evidence of this broad conservative shift. But we are unimpressed, for two reasons.
First, this ideological shift was dwarfed by shifts reflecting voters' perceptions of the economy. Steeper's 9-point anti-Democratic shift among conservatives (37 percent of the electorate) pales compared to the 25-point shift against the Democrats among those who thought the economy was not so good or poor (about 60 percent of the electorate) and the 36-point Democratic shift among those who thought their family financial situation was getting worse (about a quarter of the electorate).
Second, 1992 and 1994 exit poll data do not show a large change in the effect of conservative ideology on the House vote. More substantial shifts take place on economic variables like family financial situation, long-term wage decline, and especially assessment of the national economy. And the most important shift of all takes place in the relationship of partisanship to the House vote, particularly among independents.
Controlling for their general anti- Democratic swing, there appears to be no significant difference in the relationship between conservatism and the House vote in 1992 and 1994. What might appear to be an ideologically driven shift was really a pragmatic shift based on economic and other concerns. (See figure 1, Ideology or Economics?)
If long-term (and uninterrupted) declining living standards are indeed the basic story of 1994, making progress beyond it requires clarity on two questions: Why do such declines seem to disadvantage the Democrats uniquely? And what might the Democrats (or any other progressive political formation out there) do about that fact?
Who takes the political blame for long-term changes in economy and society depends on which story the average person believes about these changes. Here, long-term changes like steadily declining living standards differ from short-term changes in the business cycle (booms and recessions), which simply benefit (or hurt) the party on whose watch the growth (or decline) takes place. The Democrats keep getting hurt by declining living standards because the story the public believes about these long-term changes casts them as the villain. The dominant story among the general public is that the long-term decline in living standards has to do with wasteful government spending (especially on the poor, minorities, and immigrants), high taxes, inefficient and obtrusive public administration, selfish behavior by interest groups, and excessive social tolerance and loss of values. Since all of this is readily identifiable with the Democrats, as the party of activist government, poor people, minorities, liberal interest groups, and social tolerance, Democrats get the blame. Because the thrust of current Democratic strategy, especially its "New Democrat" variant, implicitly accepts this dominant story, the Democrats start every election with two strikes against them. Under extraordinary circumstances they can win. Under ordinary circumstances they tend to lose.
The only way out of this box is by offering a convincing alternative in which the government and the poor don't take most of the blame for declining living standards and in which dismantling the government isn't the main policy recommendation. Certainly this alternative includes policies to reverse wage erosion, increase job security, and rebuild communities. But perhaps most important is an alternative story that shifts the blame to other targets.
The obvious candidates for such targeting are irresponsible corporate power--the source of declining living standards and feeble government response--and its political allies. An alternative story would begin by declaring that these forces have chosen a path for the country that does not inexorably accompany economic progress (globalization, the "new economy," and so on), but does inexorably drive down living standards by eroding the material basis for family and community life. It is therefore possible and necessary to choose another path--a task from which dominant corporations and politicians have disqualified themselves by their behavior.
That such alternative paths are available is obvious from the experience of virtually all other rich nations, which have survived in the new economy with vastly better income distribution and growth patterns than our own. But this alternative account does not come easily of late to Democrats. Perhaps this is because of the fight with business interests (including Wall Street, the bond traders, and Alan Greenspan) that would be necessary to advance this alternative story and agenda. Perhaps, in our private-money system of campaign finance, it reflects the Democrats' reliance on wealthy donors. Perhaps it has to do with the current weakness of organizations that represent groups that have been losing out (for example, unions or other institutions of workplace representation).
Whatever the reasons for its current timidity, unless the Democratic Party embraces an alternative story and shows a broader willingness to contest business interests and encourage mobilization along class lines, Democrats will continue to be on the defensive and Republicans will continue to have the high ground. President Dole, anyone?o
Figure 1. Ideology or Economics?
The largest decline in Democratic support was among those who thought the economy was not good or their personal financial situation was worse, not among those who identified themselves as conservatives.
Source: 1994 VNS Exit Poll, 1992 VRS Exit Poll.
White Flight at the Polls
In 1994, whites with less education and larger wage losses were most likely to desert the Democratic Party.
% CHANGE IN DEMOCRATIC SUPPORT 1992-94
high school dropout..-11%
high school graduate.-20%
high school dropout...-5%
high school graduate.-10%
% CHANGE IN REAL HOURLY WAGE, 1979-1993
high school dropout..-26%
high school graduate.-17%
some college ........-11%
high school dropout..-12%
high school graduate..-2%
Source: Authors' analysis of Voter News Service and Voter Research and Surveys exit poll data, and Current Population Survey data.
Back to Text