by Jonathan Byron
Four years ago, I wrote a few articles before the 2012 presidential election in which I tentatively speculated about an Obama loss. At the time I thought that Romney stood a chance of winning because of the historic low levels of economic growth under Obama, as manifested in the performance of Dow Jones Industrial Average. If we were to analyse the US economy within the supply-side framework – of Jude Wanniski, Art Laffer, Robert Mundell – and use the supply-sider’s favourite economic indicator, the Dow / gold ratio (which is the value of the DJIA divided by an ounce of gold), we would see that the Dow went through an extraordinary decline from 2000 to 2012. The Dow peaked at an all time high of around 40 ounces in 1999-2001, lingered around 25 ounces during Bush’s first term, and went down sharply by the time of the 2008-2009 financial crisis. At the lowest point of the crisis – February 2009 – it reached 7.42 ounces. In August 2011, it bottomed at 6.36 ounces, and in October 2012, which was the last month before the 2012 election, it stood at 7.62 ounces. From this we could conclude that Obama’s stimulus, Affordable Health Care Act, etc., didn’t lead to an appreciable recovery after 2009, and may have even made the situation worse. Obama would lose re-election, then, and meet the same fate as Carter in 1980, and Romney would win: that to me seemed a reasonable prospect.
Here we must digress, and look at Wanniski’s political theories. The late Wanniski’s The Way the World Works (1978) gave us, not only an economic but a political model. In this classic work he posits that, at election time, the electorate always gets it right – that it will choose the best candidate, or the least worst one, at election time and shows a wisdom which is greater as a whole than it is in its parts (Wanniski here in this book anticipated the ‘wisdom of crowds’ theory). Likewise, the stock market indice, given all the information available, always gets it right – that is, it gets the value of stocks right. On the day of the stock market crash of October 19th, 1987, investors displayed the wisdom of crowds and got it right on the value of the Dow on that particular Monday – given the information available on US fiscal and monetary policy at the time. One individual investor on the day of October the 19th may have taken a sanguine view of the state of the world, and started that Monday by buying up or holding on to stocks, but the aggregate sum of investors took a different and more correct view. They voted, so to speak, and got it right. As it is for the markets, so it is for the elections: the collective knows best, and knows precisely the worth, or value, of a candidate.
It should be noted, however, that in Wanniski’s model, we in the liberal democracies don’t live in a Panglossian ‘best of all possible worlds’ in which the political system delivers perfect results every time. The electorate may choose the right candidate out of a line-up, but the candidate may not turn out well. He may, after winning office, discard some of the policies upon which he was elected, or lose touch with the electorate and misunderstand or ignore the signals it sends him; or he may simply be a mediocre, bad candidate altogether, one who was only chosen by the electorate because his opponents were just as bad if not worse. Finally, a large proportion of the electorate may decide that it’s not qualified enough to judge the issues at hand on election day and so may not show up to vote, thereby surrendering its votes to others – which is a bad thing in the eyes of some critics of liberal democracy.
So, taking all this into consideration, in 2012, the electorate made the right choice by choosing Obama over Romney. But at first sight, this doesn’t make sense. According to Wanniski, the electorate, as a whole, understands economics better than any economist, and furthermore, appreciates supply-side economics. Given this, it would to reason that the electorate would punish the candidate who has delivered mediocre economic growth and favour the candidate who has signed up to a Reaganite, supply-side platform of low taxes and stable money. But Romney lost in 2012. Why?
Wanniski’s answer would be that Romney didn’t espouse any pro-growth ideas aside from keeping the Bush tax cuts of 2003 in place (the same platform that McCain ran on in 2008). In the midst of an economic downturn to which no party offers a viable solution, the electorate, when faced with the choice between a socialist and a free marketeer, will choose the socialist. This explains why the Democrats won five elections in a row from 1932 to 1948, and why Obama won in 2008 and 2012. (Having said that, Wanniski’s model can provide, ex post, a justification for an electoral result, but it doesn’t pick winners; Wanniski himself sided with the losing candidate a few times – in 1996, he voted for Bob Dole, in 2004, John Kerry. We don’t know, then, from Wanniski, which candidate will win the 2016 election).
From a Wanniskian viewpoint, the present contest revolves around one question: how socialist, how left, is America willing to go? The Democratic Party, at least since 2008, has been hijacked by Marxists, and its 2016 presidential platform was authored by Marxists. Clinton’s own platform shows the influence of that ultra-leftism, and she is proposing the most brutal and confiscatory tax regime since the days of Hoover and Roosevelt. Hilary wants to raise the estate tax from 40% to 65% (whereas Trump will abolish it and replace it with a 20% tax on capital gains); she also wants tax hikes on income and capital gains which far exceed those introduced by Obama in 2013. Evidently she believes that this anti-growth and redistributive policy will be a real vote winner. Those who are hoping for a Trump victory beg to differ: the American people may have veered sharply to the left in 2008 and 2012, but they are not essentially left.
Many predictive models – which, unlike Wanniski’s political model, claim to predict the outcome of elections – forecast a Clinton victory this year, by a small or large margin; only a handful – including Helmut Norpoth’s, which has attracted much scepticism and derision – give the election to Trump. I favour Norpoth’s model for two reasons: one is that it can be retrofitted every presidential election since 1912 (unlike, for instance, Allan Lichtman’s 13 keys model, which only goes back to 1984) and correctly predicts the outcome of each (with the exception of 1960 – the Norpoth model forecasts a Nixon victory); the second is that it relies on votes, not on polls and GDP and CPI statistics.
Norpoth’s primary model starts with a constant – the mean of all the Democratic Party percentages of the two-party presidential vote since 1912. We add to this the Democratic percentage of the two-party vote from the last presidential election (2012) and subtract the second last (2008). Then we add Hilary’s performance in the New Hampshire and South Carolina primaries (minus the mean of the Democratic Party’s percentage of the two-party presidential vote) and subtract Trump’s performance in the same primaries (minus the mean of the Republican Party’s percentage of the two-party presidential vote), and there you have it. If the Democratic Party percentage of the two-party popular vote falls below 50% after all that, then the Democrat candidate will lose. (The emphasis lies on the two-party vote, not the popular vote: for example, Bill Clinton won only 43% of all votes cast in 1992, but 53.4% of the two-party vote; likewise, he won 49.2% of all votes in 1992, 54.7% of the two-party). The exception is 2000, when the Democrats, as predicted by Norpoth’s model, won the two-party vote by the slenderest of margins but, as was not predicted, lost the electoral college vote.
The mathematics of Norpoth’s prediction, simplified, give us these numbers: 0.9 (Hilary’s primary score) – 2.6 (Trump’s primary score) + 18.8 (Democrat vote in 2012 election) – 20.2 (Democrat vote in 2008 election) + 50.6 (Democrat average percentage of the two-party presidential vote) = Clinton will win 47.5% of the two party in the 2016 election, meaning Trump wins 52.5%.
In Norpoth’s model, a candidate’s success is determined by their performance in the New Hampshire primary. (Before 1952, the parties didn’t allow voters to vote in the New Hampshire primary, or any primary for that matter, so for elections before then Norpoth uses the sum of all votes at the party convention). For the 2016 election, he adds the performance in the South Carolina primary. His justification for this is that South Carolina possesses a large number of Afro-American voters and generally indicates how well a candidate will do in the primaries in the South. It turns out that a winning candidate may do poorly in New Hampshire and well in South Carolina: this explains Clinton’s percentage of the two-party vote in 1992, Bush 43’s in 2000 and Obama’s in 2008. In general, we should take the South Carolina primary’s results into consideration during elections – such as 2016 – when both candidates are running a primary or caucus there.
The ‘third-term curse’ plays a role as well. Incumbent parties rarely win a third term. When they do, it’s only because they won the second term in a landslide. The Republicans and Democrats won third terms in 1988 and 1940 respectively on the back of huge wins in 1936 and 1984 (Roosevelt won 60.8% of all votes cast and 46 out of 48 states, Reagan, 58.8% and 49 out of 50). But what if a party was re-elected for a second term by a smaller margin than their first? That counts against them in Norpoth’s model, as we can see from the above. The Democrats won in 2012 by less than in 2012, and as we see from Norpoth’s formula, this has the effect of lowering the Democrat percentage of the two-party vote in 2016.
The essential thing is to consider the primary result and the electoral result together. Looking at 1976, we should say that Ford should have won because of Nixon’s landslide victory (60.7% of all votes cast and 49 out of 50 states) in the 1972 election. But Ford won the narrowest of victories – around 1500 votes – over Ronald Reagan in the New Hampshire primary, whereas Carter’s victory over Morris Udall was somewhat more, as we see from the table below:
In many elections, the incumbent will easily win another term. As we can see from the above, this happens when the incumbent cruises through the primaries and faces little to no opposition, especially in New Hampshire: look at Eisenhower in 1956, Johnson in 1964, Nixon in 1972, Reagan in 1984, Clinton in 1996, Obama in 2012. The incumbent in those election years won nomination with no trouble, while the opposition candidate, more often than not, had to struggle to get his. In such a scenario, the incumbent always wins. Clouds appear on the horizon for the incumbent party only when a battle occurs for the nomination: see the Democrats in 1968, the Republicans in 1976, the Democrats in 1980, the Republicans in 1992, the Democrats in 2000, the Republicans in 2008. In those years, the incumbent party went on to lose the election.
How accurate is the model? Below we find a table of forecast versus actual Democratic Party votes since 1912:
We see that while Norpoth accurately predicts the Democrat win or loss for every election (except for 1960, which, as we know, was a controversial one), he gets the actual percentage wrong, sometimes by a wide margin. For 2008, he predicted that Obama would get 50.1% of the two-party vote – Obama got 53.7%; for 2012, 53.2% for Obama – Obama got 52%. In my opinion, the reason why he got 2008 wrong by such a wide margin – 3.6% – was that, for 2008, he only used Obama and McCain’s primary score in New Hampshire. If we are to add the average of their performance in both New Hampshire and South Carolina, the estimate of Obama’s percentage of the two-party vote goes up to 51.6% and the deviation from the actual result goes down to 2.1%.
At present, the Real Clear Politics map of state polls – with no ‘toss-up’ states – gives 28 states to Trump, 22 (plus DC) to Clinton; Trump, at 266 electoral college votes, falls short of the 270 he needs (meaning that he can win Ohio, Florida, Nevada, and Iowa – states Romney lost in 2012 – and still lose the election). Assuming that Norpoth’s prediction of 47.5% for Clinton is correct, how many states will Clinton win? From the table of two-party Democrat votes above, we see that 47.5% is closest (on the Democrat side) to the 1988 and 2004 elections (assuming a deviation of 1.4%). In 2004, the Democrats won only 19 states (plus DC) and lost what are now in 2016 the Democrat strongholds of New Mexico, Colorado and Virginia. The 1988 election presents an even more dismal picture: the Democrats only won 10 states (plus DC) and lost California, most of New England and all of the Great Lakes states except Wisconsin.
A 52.5% result for Trump would (assuming the same deviation of 1.4%) resemble the Democrat victories of 1944 (36 out of 48 states) and 1948 (28 out 48 states (third-party candidate Strom Thurmond took 4 states)); also 1976 (23 out of 50 states – Carter won the electoral college vote).
From this we can conclude that 52.5% adds up, on average, to 32 states. Let’s be conservative and say that – just like in 2004 (the last presidential election won by a Republican) – the Republican candidate wins 31 states, which is an improvement of three on the 28 Trump has now. And, for the sake of argument, we’ll give him three of the states where (according to the most recent polls) Clinton’s lead is the most vulnerable: Colorado, New Jersey and Michigan. That would add up to a total of 305 electoral college votes, a figure which nears the 286 Bush 43 won in 2004. (I’ll note in passing that some of these states are under polled: New Mexico, a Democrat bastion, hasn’t been polled since May. Clinton’s lead there could have narrowed since, and it’s not inconceivable that Trump could win it as Bush 43 did in 2004).
Some interesting what-if scenarios follow from Norpoth’s model. What if Marco Rubio or Ted Cruz were running against Clinton? They would both lose, according to Norpoth. Likewise, were Bernie Sanders the Democratic nominee, he would lose by a greater margin than Clinton to Trump. Going back further, to elections past which turned out to be decisive victories or landslides for the winning candidate, we can say that in some instances the losing party wouldn’t have lost as badly as they did had they nominated their primary winner. The Republicans should have nominated Dewey in 1940, MacArthur in 1944, Warren in 1948, Lodge Jr in 1964; the Democrats, Kefauver in 1952 and 1956, Johnson in 1968, Muskie in 1972, Hart in 1984.
Those candidates either dropped out, lost badly in other states and so couldn’t win enough overall to clinch the nomination, or had the nomination taken from them at the convention and awarded to a rival. From the Norpoth model’s point of view, the Republican primaries of 2016 gave us an exciting spectacle because of the possibility that the RNC would deprive Trump of the nomination and award it to some ‘safe’ candidate – Cruz, Kasich, Rubio – and thereby, according to the model, doom the party at the election; fortunately for the party, the voters pulled it back from the brink by nominating Trump by an overwhelming majority.
Much of the model accords with three common sense notions (which don’t seem to be held by the majority of the pundits). The first of these is that incumbent parties find it hard to win a third term. The Democrats could pull of a third victory in 2016 were Obama a Roosevelt or Reagan; but sadly for the Democrats, he is not. The second common sense notion is that candidates who don’t enjoy a high level of support from the base will lose to those who do. This is borne out by a comparison of Trump and Clinton’s performances in the New Hampshire and South Carolina primaries: if Clinton loses the election, she will lose because she didn’t enthuse the base sufficiently during the primaries. The third common sense notion is that the candidate who, during a ‘change’ election, attracts the most media attention and public interest tends to be the winner: it was Carter in 1976, Reagan in 1980, Clinton in 1992 and Obama in 2008, and now it’s Trump in 2016. Norpoth’s model bears on this when it states that, in effect, the election has been decided by the time of the primaries in February. After February the public and the media know that the decision – at least at a subconscious level – has already been made, hence the enormous interest in Trump, Trump’s family and what Trump’s policies will be. They are treating him as though he already is the president.
Going back to Wanniski for a moment, it’s a Wanniskian thesis that the electorate, if it really wants Trump, will give him the states and the electoral college votes he needs to become president. The electoral college system – which is in essence a system of weighted votes – serves as a tie-breaker, as Wanniski explains. In stock market trading, some days the investors are unable to come to a consensus as to whether to buy or sell most of the stocks on the index; the market, for that day, is then characterised as ‘flat’ or ‘indecisive’. Electoral voting may show an equal level of indecisiveness. This is where the electoral college comes into play. It serves to award the presidency to one candidate over the other in the event that the electorate can’t make up its mind and give one candidate a clear, decisive majority.
Electoral voting differs from stock market investing insofar as sometimes a portion of the electorate shows a greater intensity than the rest and really, really wants a candidate (or referendum question) to succeed. The remainder of the electorate may not care as much about the matter at hand, but for the sake of keeping the peace – and coming to a consensus – may decide to give that portion of the electorate what it wants. The electoral college’s weighted votes provide us with the mechanism which gives the minority who feel more intensely about such-and-such a question, and who are underdogs who’ve been outvoted by a majority, a chance to elect the president. For instance: suppose that there are two states, A and B, with a hundred voters in each. In state A, 40 vote for Clinton, 30 for Trump, 30 for Stein, 30 for Johnson; in state B, 60 for Trump and 40 for Clinton. If the presidency were to be awarded to the candidate with the majority, Trump would win, having received 90 votes. But state A has more electoral college votes than B, and is a winner take all state, meaning that whoever gets the majority there gets all the electoral college votes. Clinton, then, accumulates the most electoral college votes and goes on to win the presidency.
In a democracy, all votes are equal, but under a weighted voting electoral college system, some are more equal than others. The Clinton campaign seems to be banking on that fact. It knows that an unenthused base won’t elect Clinton outright, so it is pinning its hopes on a narrow electoral college victory. The line of reasoning is that the electorate doesn’t like either candidate, but a portion of it dislikes Trump intensely and will make trouble unless Trump’s opponent – Clinton – is given the election via the electoral college.
The trouble with this strategy is that it could flip the other way. Trump, at present, has only 266 electoral college votes and could lose the popular vote: but suppose that the voters of Rhode Island – a state which has four electoral college votes and traditionally votes Democrat – stage a rebellion and vote for him? This is quite possible, given that, in the one poll of Rhode Island (taken in September), shows a lead for Clinton of only three points.
Why did voters – Republican and non-Republican – show such enthusiasm for Trump during the primaries? Why did they did favour him over more traditional, ‘conservative’ candidates?
The answer – and here I depart from Wanniski – lies not in Trump’s tax plan but his stance on immigration. The American electorate doesn’t want amnesty and wants to slam the door shut on immigration: hence its dethroning of House Majority leader Eric Cantor in 2014 (Republican voters ousted him in a House primary and elected an immigration restrictionist populist, David Brat, instead). Donald Trump, in the 2016 primaries, served the same function as David Brat. He was part of the anti-immigrant protest vote, and out of the 17 candidates, he had adopted the most restrictionist platform and seemed the most likely to follow through on his promises.
The reason why in 2016 Republican primary voters re-elected House Majority leader Paul Ryan, an open-borders, free-trade, globalist, supply-side technocrat, and didn’t favour his Trumpian opponent, Paul Nehlen, was that voters believe in balance. They had just made one protectionist and nativist the presidential nominee, so didn’t want to make another the House Majority leader. Besides which, they knew that it’s the executive branch which determines immigration policy anyway.
Which raises an interesting question: did the American electorate not make Romney president because of the likelihood that he would amnesty 11 million immigrants (with the support of Republicans and Democrats in the House and Senate)? An immigration restrictionist electorate in 2012 would have chosen Obama as president and relied upon a hostile House and Senate (and Supreme Court) to oppose the Obama amnesty, which is precisely what happened.
This brings us to the Democrat nominee. If the Democrats lose the presidential election this year, Clinton – or rather, the primary voters who backed her – will be blamed. Democrats had put forth a candidate who is aged, ailing, corrupt and unpopular. Clinton, having no principles of her own, had given in to the Far Left of the party and had adopted a radical leftist platform which served to make her unelectable.
Some brave souls within the Democratic Party will make this line of attack, and it’s one I happen to agree with. But doesn’t it disprove Wanniski’s thesis that voters are always right, and that they always choose the best candidate? Not at all. In the Wanniski model, voters will choose the least worst out of a set of bad candidates on offer: they can’t determine which candidates will be on offer. The Democratic Party machine rigged the primaries in favour of Clinton from the outset, as we know. Because of a fear of upsetting the Clintons, many potential candidates for the nomination never threw their hat into the ring, choosing to wait until 2020. Unserious candidates Chafee, Webb and Lessig dropped out before the primaries, while O’Malley withdrew after the Iowa caucus. Sanders’ candidacy, which continued right to the end, gave the primaries the illusion of democracy, i.e., choice, but Clinton’s command of the super-delegates – and her large support amongst Afro-American voters – made sure that the outcome was never in doubt.