skip to content

Faculty of Economics

Regional heterogeneity, short economic memory and the Covid-19 effect

A new method of forecasting the US presidential race appears to show the Democrats will win the popular vote. The forecast is tight, but indicates voters have a short memory of seismic economic events over the past few years.


The U.S. Presidential election of 2016 caught many by surprise. Almost all polls and pundits predicted a Democratic victory but, Donald Trump won the overwhelming majority of electoral votes (304 out of 538), whilst losing the popular vote by as much as 2.9 million. This misalignment has occurred only four times out of 48 elections held since 1828.

In a paper just released, ‘Regional Heterogeneity and U.S. Presidential Elections’, Hashem Pesaran and his co-author Rashad Ahmed investigate the reasons behind such misalignments and by using data over 3000 counties show how socio-economic factors and regional heterogeneity will shape the Presidential Election in 2020.

Professor Pesaran, an Emeritus Professor of Economics at Cambridge University and a Fellow of Trinity College says “The unexpected 2016 Republican victory is explained by our model. We do not rely on polls but consider many socioeconomic factors, most importantly regional heterogeneity, to account for growing disparities in voting behaviour possibly resulting from increased political polarisation.”

2020 Presidential Election Vote Shares Forecasts by Counties. Republican/Democratic wins in red/blue

He adds the key to his findings are to exploit cross-section variations. “This shows how the relationship between many socioeconomic variables and voting outcomes systematically differ across U.S. regions.”

Professor Pesaran says that “our results corroborate evidence of a ‘short-memory' among voters; what we see is that economic fluctuations in the few months prior to the election are much more powerful predictors of how the US will vote, compared to the long term story.”

The country-wide rise in unemployment and falling incomes due to Covid-19 are likely be critical for the outcome of the 2020 election. Important factors explaining voting outcomes include incumbency effects, voter turnout, local economic performance, unemployment, poverty, educational attainment, house price changes, and international competitiveness.

The authors test multiple models, all of which forecast a popular victory for the Democratic candidate, but forecast close electoral college outcomes. “Our forecasts from the regional models imply a tight electoral college outcome,” says Professor Pesaran. “Using data available through mid-October, one model forecasts the Republicans will gain 249 electoral college votes, another forecasts 270.” 270 votes are needed to win the presidential race.

“A swing in just one or two state outcomes could swing the entire election. This point re-emphasizes why the aptly named swing states are such crucial political battlegrounds.”

The full paper R. Ahmed and H. Pesaran, Regional Heterogeneity and U.S. Presidential Elections, CWPE 2092, is available at:






US Election