In June 1987 the Conservatives under Mrs. Thatcher were re‐elected with a majority of over 100 seats against Labour. They received 42.3 per cent of the total vote, the size of the majority owing much to the significant number of votes received by third parties. But it is also believed that the Conservatives' share of the vote, unchanged from 1983, reflected the performance of the macro economy. In the four years of Mrs. Thatcher's second term, output rose more than 3 per cent a year and by the time of the election inflation was below 1 per cent, interest rates were under 10 per cent, and unemployment had come back below 3 million.If Mrs. Thatcher and her Cabinet colleagues had planned an economic strategy for the next four (or five) years in order to be in a strong position to win a fourth term of office, it might have included the following factors: output growth of 2–3 per cent a year; inflation staying around 1 per cent; interest rates of 10 per cent or thereabouts; unemployment down further from the near‐3 million mark of June 1987. On our current forecast, that is, with the exception of output, the economic record of the Conservatives' third (Thatcher/Major) term. Yet the Conservatives have been running neck and neck with Labour in tile opinion polls and, barring unforeseen developments, the coming election will be extremely close, with the possibility still of either a small Conservative majority or a small Labour majority or even a hung Parliament.Why is it that, against the background of a similar economic performance in aggregate, the Conservatives have lost popularity? The arguments are complex and a full explanation would include the introduction of the community charge arid the fall of Mrs. Thatcher herself. But the economy is part of the explanation.The economic literature on Government popularity examines the state of a number of economic variables at the time of the election. At its most extreme, some believe that all a Government has to do to be re‐elected is deliver a low mortgage rate in time for the election. Other analysts have explained Government popularity in a simple regression framework, with a lagged dependent variable to capture sluggish adjustment.A weakness of this research is that it implicitly believes all the Government has to do is to ‘get it right on the night’. As long as the economy falls into place by the time of the election, re‐election is certain. This implies that the electorate both forgets and forgives, and is indifferent to the course of the economy in the previous three or four years. But it cannot be the case that the electorate evaluates only the average performance of the economy over the lifetime of a Parliament or even the most recent developments. As in any simple utility maximisation problem, it is not just the mean of the distribution that counts but also its variance. In other words the combination of late 1980s' boom and early 1990s' recession counts against the Government in a way that four or five years of steady progress would not have done. If this is correct, it may not be sufficient for the Government to deliver low inflation and interest rates and some recovery in output in time for the election; it may simply be that the electorate remains unforgiving of a five‐year track record o