As it happens in most proportional systems, the Netherlands has consistently presented a multiplicity of parties holding seats in the Parliament. However, if the 2012 election had witnessed the VVD and PvdA dominating the stage, a much more fragmented scenario emerged from this year’s election. In 2012, VVD and PvdA collected together over half of the votes (26.6% and 24.8% respectively), with the vote shares of the other major parties fluctuating between 8.0% and 10.1% (see Figure A). By contrast, no party reached the 25.0% threshold this year. The VVD won a plurality with a share of 21.3%, followed by a second tier of competitors evenly splitting the rest of the votes (see Figure B). Geert Wilders’s party, the PVV, won 13.1%, while the CDA and the D66 almost tied with 12.4% and 12.2% each. In the lower band, below the 10% threshold, both the Green and the Socialist Party gathered a 9.1%, followed by the Labor Party (PvdA) with 5.7% and a long tail of minor parties (CU, PvdD, 50PLUS and SGP). Two new parties entered the political scene this year. DENK, promoting a strong pro-immigrant agenda, managed to win 2.1% of the votes, as much as the long-established SGP. Similarly, the FvD, an anti-establishment, Eurosceptic party, gained a solid 1.8%, which will possibly translate in 2 seats in the Parliament (The Economist 2017).
Clear rearrangements are visible when comparing the current results with the election outcome of four years ago (see Figure C). Perhaps the most striking change has been the almost complete disappearance of the PvdA from the circle of top players. The Labor Party, led by the ex mayor of Amsterdam Lodewijk Asscher, lost approximately 19.1% of the votes, falling from 38 seats in the Parliament in 2012 to barely 9 seats today (The Economist 2017). The Green Party (GroenLinks) arguably came out as the biggest winner of the election. From a humble 2.3%, Jesse Klaver managed to quadruple the support for his party touching a 9.1% vote share and thus establishing a concrete presence in the left wing. The populist party PVV (Party for Freedom) performed worse than the polls predicted (see Figure D). In late February, Will Geert Wilders was expected to win with 29 seats, but he was only able to gather 13.1% of the votes at the elections in March (The Economist 2017). Nevertheless, the party still improved from 2012 and is now the second largest party in the country. Both the centrist D66 and CDA increased their share by 4%, falling between the PVV and the SP, which remained stable at approximately 9%.
I will now outline two theories of voting and assess the extent to which they can explain the election results. I will start with the economic voting hypothesis and then present the median voter theorem.
The economic voting hypothesis revolves around the idea that economic factors determine the success of party. At each election, the electorate judges the performance of the incumbent government retrospectively and then decides for or against reelection (Fiorina 1981, 26). Generally, scholars seem to agree that “as the economy worsens, the government loses votes” (Lewis-Beck 1990, 8). This is expressed in the form of a function, the VP-function, where the popularity or vote for the government is the independent variable, while economic (as well as political) conditions take the role of dependent variables (see below).
ΔC_t={a_1 Δu_t+ a_2 Δp_t+⋯}+(c_1 D_t^1+ c_2 D_t^2+⋯)+ ε_1
ΔC_t: change in government popularity or vote share at time t
Δu_t change in unemployment rate at time t
Δp_t: change in inflation at time t
Underlying this reasoning is the Responsibility Hypothesis, which assumes that voters hold the government responsible for the economic changes happening during its incumbency, and cast their votes accordingly (Lewis-Beck and Paldam 2000, 114). As shown in the formula above, voters primarily refer to two indicators when assessing the overall state of the economy: unemployment rate and inflation (Lewis-Beck 1990, 15). High levels in both factors usually lead to a decline in government support, but the degree to which this happens varies from country to country (Evans 2004, 125). Furthermore, unemployment and inflation affect parties in different ways. Left parties seem to be more affected by changes in unemployment than right parties, whose popularity depends more strongly on inflation levels (Evans 2004, 130). Finally, reward/punishment voting only works when voters are able to identify who is responsible for the economic outcome. With this said, coalition governments are believed to be less susceptible to variations in the economy due to the inability of voters to allocate responsibility (Evans 2004, 129).
Applying the economic voting hypothesis to the case in question, we first need to understand the general trends of the Dutch economy. At first glance, the picture seems brighter than it actually is. Ranked the fifth strongest economy in the world (World Economic Forum 2015), the Netherlands benefits both from high levels of employment (74% of the working-age population has jobs, compared to only 64% in France) and high income per capita (Barker, Romei and Robinson 2017). Dutch people are among the richest in Europe and wealth is evenly distributed across all social classes (ibid.). Nevertheless, taking a closer look reveals hidden flaws that are causing discontent among the population. Although employment is high, it is still 5 points below pre-crisis levels (Klein 2017), and the number of people on temporary contracts has increased sharply (Barker, Romei and Robinson 2017). According to what said above, bad employment rates should hurt left-wing parties, which is in turn what happened with the catastrophic decline of the PvdA. On the other hand, however, economic theory does not fit when considering the effect of inflation. Indeed, despite a general decline from 2012 to 2017, prices have started rising once again since last year (Khan 2017). This would suggest a drop in VVD’s popularity, which is confirmed by the lower vote share that the party won this year, but it still does not explain why the VVD would come out as the first party in the election. More generally, economic theory says that coalition government should be less affected by downfalls in the economy, as the electorate will find it difficult to attach responsibility to single parties. According to this view, support for the government should have stayed unchanged and, as the results show, this is not what happened.
Overall, even though the economic hypothesis can explain some aspects of the election results (e.g. the downfall of the Labor Party), it ultimately fails to deliver a coherent and comprehensive description of all the rearrangements that took place in Dutch politics since 2012. Still needs to be explained, for example, the rise of PVV and the jump of the Green Party, whose policies were not necessarily centered around economic issues such as inflation or employment rates. Economic theory also does not account for the good performance of centrist parties (e.g. the D66 and CDA). Moreover, even if we decide to see the long-term decline of inflation as the reason behind the VVD’s victory, it still does not make sense that the party was losing support in 2014 when inflation was touching its lowest (see polls in Figure D). The theory therefore seems incomplete and contradictory at some points. I will now illustrate the median voter theorem, which might offer more useful insights to understand the election results.
According to the median voter theorem, parties’ policies tend to converge towards the preferences of the median voter. Hotelling was the first to observe that in the US the positions of the Democratic and Republic Party were becoming increasingly closer to each other (1929, 55). Black then reinforced the idea that when decisions are made by voting and decision-makers have single-peaked preferences, the final outcome will inevitably align with the “median optimum”, lying at the center of the preference distribution (1948, 28). Downs, finally, offered a complete illustration of the theorem. In “An Economic Theory of Democracy” (1957), he assumes that voters and candidates are rational. Rationality is conceived not as the ability to be logical or free of prejudices (4). Rather, people are rational if they act efficiently, by for example minimizing input “per unit of valued output”, regardless of the nature of their goals (5). Rational individuals, therefore, are supposed to know their preferences and use a cost-benefit analysis to make decisions aimed at satisfying those preferences. It is a utility maximization process. In election settings, parties want to win office (30-31), while voters want to elect the party that they believe will provide them with the most benefits (36). Because they are rational, parties will therefore do what they can to maximize votes, while voters will vote for the parties that they perceive as being the most likely to satisfy their preferences. In other words, it can be said that voters will tend to vote for those parties that are closer to their views on a one-dimensional political spectrum (116).
Parties develop ideologically coherent sets of policies that they use as “instruments” to attract voters (Enelow and Hinich 1984, 3). A party’s political position will be determined by the spatial distribution of the electorate because it will shift towards the point where the most voters are concentrated. When voters have single-peaked preferences and are normally distributed, for instance, parties will converge in the center, where the median voter is located, as to be closer to the highest possible number of voters (Congleton 2002, 10).
In applying this model to the Dutch elections, two considerations must be made. First, as it will become clear, the model fits only if we assume that the distribution of voters is skewed to the right. Second, we conceive the left-right spectrum in broad terms, as it is commonly referred to in left-right politics. Therefore the left wing comprises ideas such as progress, equality and internationalism, while the right wing focuses on economic freedom, nationalism and tradition (Heywood 2015, 119).
In a right-skewed distribution, the median voter is no longer placed in the center together with the mean voter, but is shifted to the right instead. According to the median voter theorem, parties must converge around the median voter if they want to maximize votes and be elected to office. As shown in the polls (Figure D), the far-right PVV was consistently ahead of all the other parties during much of the political campaign. According to the model, this makes sense as it can be said that the PVV’s right-wing rhetoric was more in line with the preferences of the right-skewed median voter. The victory of the VVD, however, can only be explained if the party had shifted to the right so as to recover the votes lost to the PVV. And this exactly what happened. The party took an increasingly nationalistic stand, promising the protection of Christian principles, drawing a line between the “real Dutch” and the rest, and threatening to “revoke the nationality of people involved in terrorist organizations” (Pieters 2017). The CDA also moved to the right, building its rhetoric on the protection of the Dutch identity (Sterling 2017) and proposing a “ban on foreign funding to mosques” (Pieters 2017). The CDA ended as the third party with the highest share of votes. Political shifts also happened on the other side of the spectrum. The vacuum left by the collapse of the PvdA was quickly filled by the D66 and GL, both offering more pro-working class policies addressing issues such as education, inequality and tax evasion (Pieters 2017, Boeffy 2017). It can also be said that because the CDA shifted to the right, the D66 was able to win the support of the centrist voters. For a better understanding of the political shifts, see Figure E below.
Nevertheless, it could also be said that the above analysis is a mere attempt of adjusting the theory to the results. As previously mentioned, the theorem does not fit if we consider a simple normal distribution, and the analysis presented cannot explain why the distribution would be skewed to the right in the first place (although this is a plausible assumption considering the incredible rise that the PVV had in the past year). Moreover, the theorem might be simplifying the election dynamics by locating all parties on a one-dimensional space. The rise of the PVV cannot be easily explained by saying that the Dutch population has gone “more right”. Some of the supporters of PVV were previously voters of the PvdA, for example, and this cannot be reflected in the graph proposed in Figure E.
In sum, I have presented the results of this year election and compared them to the results from 2012. I have then outlined the economic voting hypothesis and concluded that it is not a good theoretical tool to understand the results. I have finally turned to the median voter theorem which offers a plausible explanation if we consider a right-skewed distribution. Further research on the distribution of voters in the Netherlands might confirm or disconfirm the explanatory power of the theory.