to my research blog where I occasionally post results from the projects
on which I'm working. This is a quick and informal way to get
acquainted with -- or keep up with -- my research. I rarely
discuss more than one or two findings from each paper or project, so if
you're interested in more details, it's best to download the complete
paper, which I archive here
at Research Gate. If you just want a
list of my papers and publications, you'll find that on my cv and research
May 2014 -- Does a Luxury Goods Model of Voting Explain Why the Left Fares So Poorly in Recessions? Despite expectations to the contrary, left-wing governments performed poorly in elections during the Great Recession. Cassandra Grafstrom and I argue that this is no "one-off" but rather a systematic empirical regularity. Using two large individual-level cross-national panels we show that left governments are punished more than their counterparts on the right for economic downturns. A few scattered, and sometimes contradictory, literatures have claimed that governing party partisanship matters for electoral accountability. Survey experiments that we ran in Berlin and Michigan suggest that a luxury goods model of voting best explains voters partisan reactions to the economy. Voters associate left-wing parties with socially desirable but lower priority "luxury goods" policies that they do not support funding in hard times. The figure on the left show some expermental results -- the incumbent party vote loss from a recession treatment for differnt types of governing parties. Left incumbents are punished more than the average for all parties (the lower horizontal line). When they propose basic leftist policies (e.g. unemployment insurance), however, they are punished less than when they espouse luxury goods policies (e.g., senior centers, environmental programs). Right incumbents competing against left parties associated with basic policies receive the average punishment for a bad economy but right incumbents fortunate enough to face a left opposition associated with luxury goods policies, can even expect to gain votes during a recession. Rousseau said it best, "It is too difficult to think nobly when one thinks only of making a living."
January 2014 -- A Cross-National Measure of Electoral Risk. Believe it or not, there is no valid cross-national measure of electoral competitiveness in political science. There are lots of single-country measures such as the Ranney Index in the US or simple vote margins but designing a measure that takes account of vote volatility and isn't simply driven by national differences in the effective number of parties has proven prohibitively difficult. Rene Lindstaedt and I think we've cracked the problem. Unlike previous efforts, we also have a measure specific to the governing party, which is the unit of analysis that matters if your care about policy implications. So what is this measure of "electoral risk"? Based on (lots and lots of) district-level election data (in SMD countries) and the degree of electoral volatility in previous elections, we estimate the probability of a sufficiently large electoral swing occurring to deprive the ruling party (whether in a coalition or not) of its parliamentary plurality in the next election. We expect that leaders at middling levels of electoral risk to be most responsive, while lead parties with little chance of retaining office or little chance of losing office will have more leeway to ignore voters' concerns. Our idea here is for our new measure to support new research in political economy, representation and accountability. Some nice patterns pop out of our measure, for example the marked difference in electoral risk between first-past-the-post (SMD) systems and proportional electoral systems. The plot on the right shows that safest lead parties inhabit proportional systems, with Sweden leading the pack.
|October 2013 -- A Benchmarking Forecast of the 2013
About two months ahead of the 23 September 2013 German federal elections, Arndt Leininger and I forecast that the ruling coalition (the CDU/CSU and FDP) would get 47.05% of the vote. We got a fair amount of attention in the German press -- for example, here, here, or here -- and it turns out that we weren't too far off. The coalition ended up with 46.3% of the vote, making us the second most accurate forecast (Jerome, Jerome-Speziare & Lewis-Beck predicted 47%) and slightly more accurate than any of the many polls conducted as late as two days ahead of the election. The distinguishing feature of our forecast model is that it uses the deviation in German growth from an international benchmark (French, UK and Italian growth) instead of just growth. In most years this doesn't matter much but in 2013 German growth deviated considerably from that of its peers and the press portrayed the economy positively, despite objectively modest growth figures. See my 2012 Benchmarking across Borders paper with Michael Peress for the full logic behind this. One part of our forcast that didn't go very well, however, was our addendum that given their vote share, the coalition had an 83% percent probability of remaining in office. The FDP tripping on the 5% hurdle to get into parliament hurt us here.
June 2013 -- How Stable is the Economic Vote? Really. I was asked to write a chapter for the next edition of the LeDuc, Niemi and Norris book, Comparing Democracies, and thought this would be a nice opportunity to drag out some truths on the economic vote that would be difficult to get into a journal. Everyone who works on the the relationship between the economy and the vote knows that the relationship, especially at the aggregate level, is extremely unstable but from reading the journals one would think that it is as sound a relationship as suppy and demand (milk and cookies? you get my point). The book is intended as an introduction for students and non-specialists, so I thought it would be fun to trot out what the basic relationship looks like before things get complicated. Results like these are no surprise to anyone who works in the area and serve as a jumping-off point for a lot of the literature that seeks to improve estimation of the vote. The bizarre thing is that the weak results don't appear in journals because they don't show much of a relationship. The plot on the right shows the simple bivariate relationship between unemployment and the vote share of the lead party in the government in 23 developed democracies for election years with data between 1955 and 2009. Economic accountability leads us to expect incumbent vote share to drop as unemployment increases. In these simply aggregate level plots, only 13 of the 23 countries even show a negative relationship! Using growth instead of unemployment doesn't improve things much. Of course, great work has been done showing that the results improve when conditioning for institutional and political context, or when individual-level survey data with economic perceptions is used, or when local instead of national economic data are used and many more variants. What you see on the right, however, shows why this endeavor is not always so easy!
July 2012 -- Do Voters Really Hold Governments Less Accountable under Globalization?
The most imporant finding in mass political economy in recent years, in my humble opinion, is that the effect of the economy on the vote diminishes in more economically open settings. Tim Hellwig had the original finding on this in 2001 and this relationship underpinned a large part of the most important book on the economic vote in recent years, Duch and Stevenson's appropriately named, The Economic Vote. But do voters really hold governments less accountable under globalization? It sounds reasonable that voters might understand that governments have less influence over open economies in which shocks are more likely to come from abroad. Such reasoning may demand an unlikely degree of sophistication from voters, however, and as Jack Vowles points out, surveys don't seem to show that voters in open settings think their governments are less efficatious. Michael Peress and I, as part of a proposed CSES edited volume that includes Jack Vowles and Tim Hellwig, among others, offer an alternative explanation for this empirical regularity. Economic performance in open countries deviates less from the international business cycle and this deviation (Local Growth in the figure at left) drives the economic vote, as we show in our 2012 Benchmarking across Borders article. If this is the case, then voters are likely just as responsive to the economy as they've ever been; the smaller economic deviations in open economies are what have attenuated the economic vote.
|April 2012 - Fairytale Trust?
Most everyone believes that culture has an important effect on
political and economic behavior but only a small proportion of
empirical political science and economics even uses culture as a
control variable, let alone a theoretically imporant explanatory
variable. There's a simple explanation for this: culture is
difficult to measure and those (mostly survey) measures that do exist
are likely endogenous to the measures they are supposed to be
predicting. What researchers need is an exogenous cross-national
measure of culture, something older that is not influenced by
contemporary variables, something like ... 19th century
fairytales. Together with a team of seven research assistants, I
coded and average of 22 fairytales from over 30 countries on a large
number of cultural characteristics patterned on the World Values
Survey. One team of RAs collected and anonymized the stories,
replacing all indicators of national origin such as names, food, or
places with anglicized equivalents. The second team read and
coded the anonymized tales. The results for questions related to
trust and collectivism pass a casual plausibility test and even
correlate reasonably highly (.5) with comtemporary cross national
cultural survey results. Shanker Satyanath and Dani Marinova
and I are now preparing to apply for grants to expand the data and to
apply it to substantively important problems, such as estimating the
cultural (trust and collectivism) effect of communism in Eastern
Europe, the cultural colonial origins of development and other
topics such as tax compliance, corruption and quality of government.
|February 2012 -- Benchmarking across Borders.
So what does 1.7% growth mean? Apparently not so much without
context. Consider a lesson from Donald Tusk, the former and current
prime minister of Poland. In the run-up to the 2011 general
election, the Polish economy was underperforming its recent rapid
growth but it nevertheless outstripped that of its European neighbors
struggling through the financial and Euro crises. This point wasn't
lost on Tusk who declared Poland to be a "green island" and projected a
map with national growth rates behind him during public appearances. He
went on to become not only the first prime minister in Poland but
in all of Eastern Europe to be reelected since the fall of
communism. Donald Tusk's triumph is a particularly clear example
of the type of cross-national benchmarking that forms perceptions of
the economy that voters then use to evaluate governments. In our
2012 APSR article, Bencharking across Borders, Michael Peress
and I show in a large sample of developed countries that economic
performance relative to that of other countries matters more than
"non-benchmarked" performance for the electoral fortunes of lead
parties. It is not that voters themselves compare growth or
unemployment across borders but, as we show using a content analysis of
the Times of London, the media report more positively on the economy
when it's doing better than the neighbors'. Our explanation of
the economic vote also sheds light onto another puzzle, why so many
incumbent parties that came up for reelection during the Global
Financial Crisis managed reelection despite dismal economic
performance: if releative performance is what matters, then not all
countries can fare badly. This is a huge topic and we've
continued chisel away at it. We're now completing the collection
and analysis of over 2 million newspaper reports about the economy in
35 newspapers in 17 countries since as early as the late 1970s.
We hope to identify the countries that are used as benchmarks for
other countries, convincingly establish that the media benchmarks
across borders, show that the reported economy matters more than the
objective economy for the vote, and examine whether papers on the left
and right benchmark similarly. You'll be hearing more on this!