« Organic Milk and Japanese Cars | Main | Statistics Canada under siege »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Perhaps of related interest:-


A variant of #1 inspired by Krugman's latest observations: Conservatives tend to be elected when people think they want to shrink government. Then when GDP shrinks and services are cut, people get annoyed and toss them out.

For whatever it's worth, there's a closely-analogous pattern in the US.

Since Franklin Roosevelt took office in 1933, every period of Democratic administration has shown better economic growth than any period of Republican administration. It's really quite striking ... I've often wondered whether there's causality, on merely correlation

Roosevelt/Truman 3.8%
Eisenhower 1.2%
Kennedy/Johnson 3.5%
Nixon/Ford 1.7%
Carter 2.3%
Reagan/GHWBush 1.9%
Clinton 2.5%
GWBush 1.3%

% growth p.a. in real GDP per capita, from the last year of one administration to the last year of the next

Bill, Patrick, Passing By - thanks for those links, and for your explanation Patrick. Are there other explanations out there of the Democrat growth premium? Does a similar phenomenon exist in other countries?

Bill - as regular readers know, I'm not at all a macro person. But that stylized fact - the federal debt grows faster under conservative governments - holds in Canada for the past 25 years. The debt grew under (Progressive Conservative) Mulroney, feel under (Liberal) Chretien and Martin, and has grown again under (Conservative) Harper. Before then the pattern is not so clear - the debt grew substantially during the Liberal governments of Pierre Trudeau and - if one includes war time debt - MacKenzie King. So I don't know what the pattern would be if one looked at all the years since Confederation.

Third hypothesis: Conservatives appoint inflation hawks to the BoC.

Okay, I don't buy that one, but John Crow would be one data point supporting that point. But Coyne-Diefenbaker spat would be one in the opposite direction.

This relates to what Passing By noted above.

Larry M. Bartels of Princeton showed in “Unequal Democracy” that the U.S. economy at least since 1948 has performed better under Democratic than Republican presidents for all income classes, but under Democratic presidents income inequality tends to decline. He attributes it more to differences in public policy than any other cause. Here is a good NYT article that serves as a review of his book:

“Such a large historical gap in economic performance between the two parties is rather surprising, because presidents have limited leverage over the nation’s economy. Most economists will tell you that Federal Reserve policy and oil prices, to name just two influences, are far more powerful than fiscal policy. Furthermore, as those mutual fund prospectuses constantly warn us, past results are no guarantee of future performance. But statistical regularities, like facts, are stubborn things. You bet against them at your peril.”


The fun thing is ask any Chamber-of-Commerce type and they will tell you how much better the Repug-Conservatives are good for business. Or maybe they just mean that they are good at increasing inequality, which, for some, is obvouisly a good in itself.

I think the data set is just not significant enough to warrant drawing any conclusions from. 135 years sounds like a long time, but that's really only 10 to 15 macroeconomic cycles. In the same time, Canada has only switched from a government of one party to a government of the opposite party 16 times (the total number of Prime Ministers is irrelevant, since we're concerned only with parties). Those aren't very compelling figures upon which to base conclusions, and in that time Conservatives have governed through four global economic crises that can't possibly have been the result of their domestic policies (the Long Depression of 1873-1896, The Great Depression of 1829-1939, Black Monday 1987 and the subsequent late 80s/early 90s recession, and the Financial Crisis of 2007-2010); surely that will skew the results.

In any event, I'm inclined to dismiss an explanation qua partisan economic policy out-of-hand - most governments spend very little time in power during the time in which their macroeconomic policies will have the most effect. If I were to do this study, I would first correct the data set to mitigate the effect of global economic factors, then I would only look at the economic performance of a party after a certain number of years following their assumption of power (say, four years). Then you could draw some conclusions about how each party effects the economy. My hunch is that you might be able to find something about specific long Prime Ministerships (i.e. the effect of Trudeau's policies on the performance of the Canadian economy in the late 1980s), but very little about the parties in general.

Geoff NoNick: " Conservatives have governed through four global economic crises that can't possibly have been the result of their domestic policies (the Long Depression of 1873-1896, The Great Depression of 1829-1939, Black Monday 1987 and the subsequent late 80s/early 90s recession, and the Financial Crisis of 2007-2010); surely that will skew the results"

* your last two data points (Black Monday, Financial Crisis) are consistent with the idea that Conservatives get voted in when people are feeling rich, therefore are in office when the bubble bursts.
* the Tories did get World War I economic growth, and they were also out of office in 1929. and the Liberals got a bit of that 19th century Long Depression. So it's not that the Tories always get bad luck.
* the positive coefficient on Liberal governance is pretty robust - it comes through in all of the specifications, and still appears when the data is chopped into different time periods. (pre and post WWII, 1960 onwards, etc).

I think Geoff probably hits it right on the head, the dataset just isn't large enough to be meaningful. I've seen similar (albeit less scholarly) claims (http://www.nytimes.com/interactive/2008/10/14/opinion/20081014_OPCHART.html?ei=5070&emc) with respect to stock market performances under democrat and republican presidents, in which the democrats significantly outperform the republicans.

Even if you accept the premise that presidents can drive stock market performance, when you take a look at the underlying numbers and it becomes pretty suspect. The republicans are pulled down by Hoover (1929 crash) and Bush II (dot.com crash) and the democrats are pulled up by Clinton (dot.com bubble). Take out those three presidents, and the whole story disappears.

Go through the same story in Canada. We've had 40 parliaments (technically you can have more than one government per parliament and there have been some short-lived caretaker governments, but let's ignore them and assume 1 government per parliament) since 1867. 16 of them had conservative governments (there was also 1 coalition government, but let's ignore that) of some sort or another (although 5 were relatively short-lived - the current minority has lasted longer the McDonald's 1872 majority government). Of those 16 governments, 5 overlapped with the long depression, 1 overlapped with the great depression, 1 overlapped with Black Monday, 1 overlapped with the recession of the late 80's/90's, 1 (and maybe 2, depending on when you think it started) overlapped with the 2008 crash. So we have 16 conservative governments, 9 or 10 which overlapped with negative, and clearly exogenous, adverse economic events. Moreover, the remaining 6 or 7 governments include 3 minority governments which lasted a year or less (Dief twice and Clark).

Now, one could tell a similar sob story with respect to the Grits, Trudeau had stagflation in the 70's, King was in power in 1929, McKenzie had to deal with the long depression. But in many respects the Liberals appear to lucked out in terms of when they game to office. The modern Liberal dominance in Ottawa started in 1896, coinciding with the end of the long depression (and the Wheat boom in Canada). King may have led Canada into the great depression, but he also led Canada out (and through WWII), St.Laurent managed to ride the early years of the post-war boom and, after the Dief interregnum, Pearson and Trudeau benefited from the tail end of those years, and Chretien and Martin took over just as Canada was emerging from the recession of the early 90's. While the Liberals had their share of bad luck, for the most part, they've been significantly luckier than the Tories.

I think they're just not properly controlling for exogenous fluctuations. On that note, why did the authors use US economic activity to control for exogenous economic fluctuations? I mean, that's probably a reasonable control in 2008, but I would have thought that, prior to WWII (and certainly prior to WWI) the UK might have been more significant for Canada's economy and a better proxy for global fluctuations (query whether that might distort the pre-WW II analysis).

This is why it pays to know your history. First, economic growth differentials during the McDonald/Laurier eras can be explained by one thing: The United States. Both McDonald and Laurier pursued the same National Policy, build a transcontinental railway (CPR for McDonald, Grand Trunk Pacific and Canadian Northern for Laurier), protect Ontario and Quebec manufacturing with tariffs and populate the West with homesteaders.

The US pursued the same policy during McDonald's time and thus Canada couldn't attract enough Western settlers, they went to the US instead. The CPR nearly went bankrupt due to this. Laurier carried the very same policy but the US was full and getting out of the homestead business and thus his policy succeeded as he managed to get the West settled.

I would also like to point at that the Tories and the Liberals crossed the road over Free Trade in the late 1970's. Laurier campaigned on Reciprocity, er, Free Trade with the United States in 1891 and 1911. The Liberals lost the 1911 Election over their support for Reciprocity. At this time the Liberals were the Continentalist Party and very much in favour of Free Trade.

We as a nation revisited this very question again in 1988.

Secondly, we didn't have a Central Bank until 1935. Before then the Bank of Montreal was the fiscal agent for the Government of Canada, managed the National Debt and carried weight as the "Senior Bank". It was an era of free banking which makes Canada a lovely example to the rabid free-banking fringe.

Third, Canada's government response to the Great Depression was quite different than the Americans. Bennett took a hard line until 1935 until when faced with almost certain defeat he had a change of heart. "Bennett's New Deal" See http://en.wikipedia.org/wiki/Great_Depression_in_Canada#New_Deal for details. Most of the legislation was thrown out by the courts, only the Bank of Canada and the CBC survive. Unlike the New Deal in the United States which was a genuine change of policy, Bennett's policies were rejected by voters as nothing more than a cynical attempt to bribe them. Mackenzie-King during the Depression was a little kinder than Bennett but didn't enact anything approaching the New Deal.

The difference with Mackenzie-King was that unlike Churchill and Roosevelt he hung on in power until 1948. The roots of the leftist Liberals we know date from this time. The CCF, predecessor to the NDP started to win seats and Tommy Douglas came to power in Saskatchewan in 1944. The Liberals were scared and essentially stole most of the CCF's popular programs. Baby Bonus, Veteran's Charter, Employment Insurance, the Welfare State as we know it. The modern Liberal Party really dates from this time when they leaned seriously Left.

So, after saying all that, my conclusion is that graphs like Frances are fun to look at but they are essentially meaningless. Parties flip-flop on policies and policy directions with regularity. They steal their competitors ideas and make them their own. They repudiate their earlier policies. It's an essential form of dialogue that makes government work, but it also makes a mockery of claims that "GDP is higher under Liberal Governments". Correlation is not causation.

Determinant, Bob - given that the same pattern is found in the US, and the effect is large and found over a long period of time, just to say that it's "essentially meaningless" seems a bit strong. I could see the "essentially meaningless" interpretation if the effect was, say, 0.05 or 0.5, but when the estimated coefficients are 1.6 to 2.3, it seems that some kind of explanation is called for.

Now I can see the argument that the causation runs from economics to elections - i.e. Conservatives tend to get elected either just before or just after the economy goes in the tank.

Although Ferris can point to a smoking gun: monetary policy is more expansionary under Liberal governments.

It's a matter of balancing growth in supply with growth in demand: Say business has a nominal rate of profit of 6%. With 0% inflation, it could reinvest and grow 6%, but what would be the point? Business, that is supply, would grow but demand would not. Suppose instead you allow 3% inflation. Then business still gets its nominal 6%, but really only gets 3% profit. Demand also grows 3%, as everybody gets their cost of living raises, and effectively shares in the real profits of business.

Inflation is, among other things, a redistributionalist tax on business profits. It distributes demand to the rest of the economy, where otherwise business would retain the entire increase.

Note that now, with inflation 'tamed,' businesses are awash with profits, but still refuse to invest. Why should they, when demand has not grown? And under the circumstances, will not unless they hire, or export, more.

That's why there is more growth under expansionary monetary policy, thus under Liberal governments. It also shows that the Conservative tendency to 'keep everything to themselves,' is counterproductive.

Thanks. You read it here first.

Hi Frances,

I got into this deep enough to go have a quick look at their intro and think a bit. Now, they do acknowledge from the outset that causality can go either direction (business cycle to politics or vice-versa). Furthermore, I think you would already need - at the very least - a theory of how the business cycles would have been going on, without state intervention, in order to have some grip on that causality. I'd be extremely skeptical that this is really empirically feasible. However, I think a kind of public choice type story is perfectly compatible with your hypothesis number 2, although I'm not sure I buy the pro-poor bit.

If people are sophisticated enough to have a general concept of economic cycles, in other words, that they know things will probably get better when at the bottom and worse when at the top, and if people are generally risk-averse (despite buying lottery tickets?), they'd anticipate a reasonable chance of getting a bigger piece of pie if they put an expansionist government in power during the upturn, and a very large risk of getting an even smaller one if they did the same for the downturn. There are many ways that you can redistribute the increase in size of a pie without hurting anyone, but the more you play around with a shrinking pie, the more someone is bound to get hurt.

I suppose it may be just a verbal twist on your pro-poor interpretation, but I don't think you have to *feel* poor in order to want a bigger piece of pie, or *feel* rich to fear losses. But you may switch perspective based on your expectations, and this is what I see as the effect of the cycle on politics. Otherwise, one frankly wonders why anyone ever voted conservative... Don't we all want the world and want it now?

Frances: "Given that the same pattern is found in the US, and the effect is large and found over a long period of time, just to say that it's "essentially meaningless" seems a bit strong."

I bet that if you ran the same model using US economic growth instead of Canadian economic growth, you'd end up with similar parameters for the Liberals and the Conservatives - even though there's no plausible story about Canadian governments driving US economic growth.

I know what you're saying (that its the same pattern with Republicans and Democrats), but it's the same problem in the US as it is here, there just aren't enough datapoints to be able to distinguish party factors from exogenous economic factors. Maybe in a few hundred years.

Bob. but there *is* a plausible story about economic growth driving the choice of government, i.e. people vote for parties that offer social insurance when they think they're likely to need it.

This is slightly off-topic, but I think it *might* be related.

IIRC, the US stock market does better under Democrats than under Republicans. Does this mean Democrats are good for stocks? Not necessarily.

Assume Democrats cause lower profits. That makes Democrats bad for stocks. Suppose we have a Democratic President. As the election date approaches, there's a positive probability the next President will be Republican. The EPV of profits is rising as the election date gets closer. So stock prices are rising. And if we have a Republican president, there's a positive probability the next president will be a Democrat, and the EPV of profits is falling as the election date gets closer. If stock prices = EPV of profits, they will be rising under a Democrat president and falling under a Republican.

That's easier to see if you assume the two parties take turns to govern. Assume $1 profit under Republicans, and $0 under Democrats. At the beginning of a Democrat regime, the flow of profits is $0,$0,$0,$0,$1,$1, etc. At the end of the Democrat's Term the flow of profits is $1,$1,$1,$1,$0,$0, etc. The PV of that flow is higher at the end of the Democrats term than at the beginning. Asset prices are forward-looking.

The whole economy is one big asset price. That's why the economy grows under the Liberals. Things can only get better, and the closer we are to the election, the sooner things can get better ;-)

Nick, interesting, Steve Ferris could test that idea quite easily with his data.

One related hypothesis I had for the build up of debt under conservative governments, e.g. the graphs linked to above showing higher debt growth under republicans, is that it's cheaper for republicans to run up debt than democrats, because financial markets trust them and charge them lower risk premia.

Greg, interesting, also related to the discussion a few posts ago on corporate income taxes.

Now that I think about it a little more, my earlier hypothesis seems improbable. The Liberals cut lots of services during their last 12 year run and it didn't seem to hurt them all that much. I think Nick's forward looking asset price argument improbable for similar reasons. During the last Liberal run was business looking forward 12+ years to the coming bonanza under a Harper minority?

Also, I'd say people become more mean spirited, uh conservative, as they get poorer. If you're feeling poor, are you likely to elect a political party that promises to take away more in taxes and give more to those worthless welfare queens? So the feeling rich/feeling poor arguments doesn't convince me.

The straw I'm currently grasping at is this: I could be wrong but off the top of my head it seems to me that conservatives/republicans spending more on the police state (a.k.a law and order) and defence (and wars) than liberals/democrats. What if this spending is more of a drag on growth than whatever liberals/democrats tend to throw money at (welfare queens, drug injection sites, grants for mediocre artists, etc ...)?

Patrick: "Also, I'd say people become more mean spirited, uh conservative, as they get poorer. If you're feeling poor, are you likely to elect a political party that promises to take away more in taxes and give more to those worthless welfare queens? So the feeling rich/feeling poor arguments doesn't convince me."

So that's an alternative explanation of why we'd see that pattern in the first graph - an economic decline *causes* a conservative government to be elected.

The military spending argument is interesting - and the part of military spending that tends to increase more during Conservative governments is "guns and bombs" - i.e. military hardware, helicopters, etc. And it's plausible that this particular type of military expenditure would have a fairly low multiplier.

This study is worthless. GDP is over-rated. I remember a story my dad told me, he was before retiring an economics professor at the U of A. After the USSR died Western economists got a chance to study the USSR economics model. You see, up till than they couldn't understand how its GDP was so high (think china and India of today). They realized that their concerns they had for decades was unwarranted. There is such a thing as Bad GDP growth. That's when the production does not effect anything other than some statistic, and in fact without financing to hide it cost, it costs more than it takes in.

The case study a speaker who came to the university used was a window factory. Communism does have its rich and poor. Just the way its riches are gained are different. You get awarded them rather then seek and purchase them yourself. This factory manager had his yearly quota. If he made it he was rewarded. If he went over he got more rewarded (the use of a resort cabin on the Black sea). Obviously he wants to be rich so made endless amounts of windows, which used up allot of inputs. Stuff like this makes the GDP big, you see.

He didn't care about the quality, nor did he care about anything other than making his quota. On transportation from factory to warehouse between 50-60% of windows broke and ended up in landfills (think China of today - my example: I bought from Wal-mart a water strainer made in China. After 4 uses it rusted. Junk!).

*These broken windows were not removed from GDP*. That's more likely why the GDP may be higher under the Liberals. They spend and spend making junk (whether it be services or products) we don't need. Bad GDP growth.

Stephen -"This study is worthless."

Before making a blanket condemnation of the Ferris and Voia paper, I'd suggest you take a look at it. Their interest is in political opportunism, that is, whether or not governments attempt to stimulate the economy in the lead up to an election. They certainly do not set out to show that Liberal governments generate higher economic growth. Steve Ferris is about as far from being an apologist for the Liberal party as you can imagine.

But the fact is, if you're trying to explain GDP growth over time in Canada, throwing a dummy variable for "Liberal" into your regression equation improves the fit considerably.

Frances wrote:

Bob. but there *is* a plausible story about economic growth driving the choice of government, i.e. people vote for parties that offer social insurance when they think they're likely to need it.

That goes to the root of my objection to your claim: party policies, their whole positions, even the limits of the political spectrum are not static. The Liberal Party championed Free Trade with the United States in 1892 and 1911 and lost. The Tories opposed Free Trade at that time but promoted it in 1988. Canada openly promoted being Income Tax free until WWI to attract British immigrants, including my family. We've had an Income Tax since 1917.

Neither the Liberals nor the Conservatives promised anything like Social Insurance before the 1930's. The Liberals didn't come around to the idea until 1945 when they were scared the CCF/NDP would eclipse them the way Labour did to the UK Liberal Party.

It is anachronistic to examine the party positions of the past through the lens of today.

Or to put it another way, your hypothesis reminds me of the old saw that to a man with a hammer, everything looks like a nail. Or perhaps to an economist everything looks like a graph.

Quite an interesting correlation! Especially since the U.S. GDP growth seems to be about 0.5% points better under the Democrats than under the Republicans. It would be interesting to see what the story is in other countries. :)

Frances: "Bob. but there *is* a plausible story about economic growth driving the choice of government, i.e. people vote for parties that offer social insurance when they think they're likely to need it."

In theory it's a plausible STORY, but only if you're an economist. As determinant points out, the implicit assumption that the Liberals offer "social insurance" is a debatable one even now, and was certainly not the case before WWII. Indeed, in practice, there hasn't been much of distinction between the Tories and the Grits on a fair bit of economic policy and it has often been debatable which party is further to the left than the other.

Moreover, equally debatable is that social insurance is what drives voters. The collapse of the Liberals in 1984 (from which they have never really recovered) had everything to do with typically Canadian regional resentments (over the NEP and the repatriation of the constitution) and a government that was getting long in the tooth. Arguably it's the same story for the Tories in 2006 (indeed, as an aside, it's remarkable that in the 27 years since 1984 the Tories have governed for 14 of them - that's a period of relative conservative dominance unseen since the Macdonald era, more so since for 11 of the other 13 years they were divided into two parties), and in 1957 (though, in 1957, the Liberal collapse in Quebec proved to be temporary).

Economically Diefenbaker wasn't a real right-winger. Actually, I was told by a person whose father had been in Dief's cabinet that Dief was quite the fence-sitter on pretty much everything. Sure he's remembered for his long-winded defence of the Red Ensign in the Great Flag Debate but he was in Opposition then, never to return to power.

He was responsible for the Coyne Affair. Other than that his economic record is slim and unremarkable. He cut taxes, raised old-age payments and increased pay for the Public Service. He was a populist, pure and simple who ran under the PC banner.

On the topic of Liberals and Social Insurance, I am still waiting for National Pharmacare as promised in the Liberal Red Book of 1997, floated again in 2004, 2008, still floating. I'm still diabetic guys.

Yeah, I believe Liberal promises about Social Insurance, yes indeed.

Theory: Liberals are pragmatic; Conservatives are ideological. Liberals will do whatever their research tells them will grow GDP; Conservatives will do what they think is right, whether or not they have research indicating that it will grow GDP.


1. I know nothing about Canadian politics beyond what I can guess from the names of the parties and by analogy with the US. For the US, I could go into a whole historical narrative, but I don't want to make this comment too long.

2. I think my theory is consistent with the observation (made by other commenters) that the parties have flip-flopped on particular issues. What is conservative in one era may not be conservative in another, and what is pragmatic certainly changes over time depending on the latest research and experience. But the underlying nature of liberalism and conservatism can still be consistent over time.

3. I think my theory is consistent with research indicating that liberals are more influenced by new information than conservatives are.

4. I'm not saying that liberals are 100% pragmatic and conservatives are 100% ideological. I recognize that there are plenty of exceptions on both sides, when actual parties are involved. (Not sure about Canada, but for the US, for example, the fact that there is no real leftist party means that the Democrats have to do some things to please the left and can't be entirely pragmatic. And Republicans have to respond to voters' desire for results and cant't be entirely ideological.)

5. I'd be interested to see patterns for other countries, particularly those that have more ideological left-wing parties, which could be a test of my theory.

6. My theory doesn't necessarily mean that liberals have better policies. In the US, for example (not sure how it played out in Canada), the Democrats kind of screwed up in the 60's by following their research when ideology would have done a better job.

I think your theory is not far off, Andy, at least in the present political context here in Canada. Of course that could just be my political bias.

It seems to me that the Liberals were headed in the right direction with corporate tax cuts before their defeat in 2006. They were pragmatic on GST, and on carbon pricing. They've been scared off both by superior rhetoric on the other side. I suppose the Conservatives tend to be more consistently ideological.

On the question of whether or not one can consistently identify political parties with a set of policies: perhaps not.

But one can make the case that one can consistently identify political parties with a set of *interests*, for example, the Liberal party tends to represent urban interests, the Conservatives rural ones, and so on. Yes, there are shifts in these alignments of interests - e.g. in school I was taught that the Conservative party lost the support of Quebec forever when John A MacDonald condemned Louis Riel. And when women gained the right to vote their interests became relevant to parties seeking election. But those shifts are fairly rare.

If the Conservative party has consistently represented the interests of older Canadians (the 'we all get more conservative as we get older' theory), and if older Canadians have a strong preference for low inflation but weak concern for job creation, then Conservative parties will adopt policies that are less pro-growth.

The other thing I'd like to check is causation through immigration, i.e. Liberals favour higher immigration levels, and higher immigration leads to higher GDP growth. Though it would be hard to test this as support for immigration goes down in hard economic times, and Conservatives tend to govern in hard economic times.

I suspect these results would disappear if the sample were to include provincial observations. There is a well known phenomena in the federation that provinces often do not vote in the same party at the provincial and federal levels. BC, Ontario and Quebec have serious contrarian tendencies. Given the division of powers as set our in the BNA it strikes me that provincial governments have at least as much capacity to alter growth. In short until I see a study that takes both federal and provincial into consideration I am going to say that the liberal premium is phantom of poor research design.

My take on a general policy distinction between Liberals and Conservatives over the long run, which would be consistent I think with the Dem/Rep split in the US, is a preference for concentration of power at the federal level for Lib-Dems, which implies in and of itself more federal spending. And unless the case could be made that provinces/states respond to that by cutting their own spending, I think my assumption isn't that far-fetched, that you want your piece of pork when someone else will pay (i.e. when anticipating an upturn - profit increase), and would rather have no pork for no one when you think you'll pay (ditto on the other side of the cycle).

I don't buy the pragmatic/ideological either. I just see superior rhetoric as the very hallmark of being pragmatic. And since WWII, the superior rhetoric has been social assistance (and other *minor* things, like public sector jobs, more business subsidies, stronger labor laws) when it could look rosy ahead, and just plain taxcutting when it didn't look so good.

That's only true after WWII, Yvan. Sir John A. McDonald was a centralizer without equal, Laurier was just a bit behind.

Until the 1930's the main job of the federal government was to guide and build a transcontinental railway, populate the West and build up Eastern manufacturing. When that was done by the 1920's Ottawa started to run consistent budgetary surpluses. It is very apparent that the men sitting at the table in Quebec in 1864 did not anticipate the Welfare State as we know it. They didn't reject it, they simply didn't think about it. Nobody in Canada did. It wasn't part of the political spectrum at that time.

The modern era of provincial/federal power arguments dates from the Rowell-Sirois Commission and the Federal takeover of the Income Tax System in 1939. Ottawa ran away with the bulk of the income tax at that time and has never returned to the prewar split where provinces actually collected most of the money.

Not quite right determinant. The feds did not run away with anything. The DOP in the BNA act gave the feds tremendous taxation power and quite a bit of responsibility via the residual powers clause. But after the privy council ruled time and again in the prov's favour by reading down and across so that feds residual powers were highly circumscribed little in the way of responsibility was left with the Feds.

And a stipulation of difference should be made between pensions for example and the way the feds got power in other areas. Prov powers were "undermined" by the Feds saying "we have a sack of cash if you want it but here is what you voluntarily have to do to get it" (Canada health act for example). That is the golden rule.

In any case it is still a lame paper. Economists and political scientists should talk more it would probably help both in research design on questions of political economy.

The same in US. Democrats always save economy from incredible budget defecit.

In response to the earlier theoretical argument (re stock exchange prices) made by Nick Rowe, I'm pretty sure that Angry Bear looked at this (for the US) in detail and came to the conclusion that it wasn't supported by the evidence - there was no evidence of prices dipping at the end of Republican administrations, no evidence of them rising at the end of Democratic administrations and no evidence of lower profits during democratic administrations. I'll try and find the link.

And I know that Paul Krugman pointed out wrt GDP growth that it seems the growth pattern was different between Democratic and Republican administrations - Republican administrations tended to raise growth (relative to Democratic administrations) in election years, but depress it the rest of the time. Don't know if the same pattern can be seen in the Canada.

I found this - not at angy bear:

"What would cause such a large difference? One possibility is that Wall Street investors expect the Democrats to be bad for the market and sell their stocks before elections that the Democratic candidate is likely to win. Then, when the Democrats do not prove as bad as expected, stock prices rise again.

The authors, though, find that the data do not support this theory - stock prices generally do not tend to decline before elections that Democrats win.

But this finding itself raises a second puzzle. If returns are so much higher for Democratic presidents than Republican ones, shouldn't we see investors rushing to the market when a Democratic victory looks likely?

We don't see that happen, either. "In sum,'' the authors write, "the markets seem to react very little, if at all, to presidential election news." "

Determinant - maybe I should specify that I didn't intend the concept of centralizing in the sense of winning tugs of war with the provinces so much as simply making the federal power relevant to voters in whatever ways conservatives in general are relatively reluctant to, given they associate with more of a *grassroots/traditional/local/non-state institutions* type of narrative. You're also right to point out that the Welfare state is a recent concept, but I've never suggested otherwise. My reading - for now anyway - of the observed correlation is that people are more willing to accept more re-distribution when good times are a-coming, yet social assistance and the welfare state is only one type of re-distribution. Building a railway certainly is one as well, as is the case with providing most public goods, and so can budget surpluses, as this just mean more money in the redistributive bank from wealthier taxpayers.

Unless I'm mistaken, the type of correlation we're looking at does not require anything like a strict adherence to a special kind of political agenda, beyond wanting governments, on average over time, doing relatively more or less stuff. And as far as I know, without being particularly knowledgeable history-wise, our political labels have built their tradition/reputation on this alternative.

With respect Yvan, I'm saying your small-government/federal power relevant simply does not fit the political legacy of John A. McDonald at all. It really was a different time. Back then the Liberals were Continentalists, wanting Free Trade with the United States. The Tories were Imperialists, wanting Canada to follow its own development path. The Tories won, that's why we got the CPR.

When I say McDonald was a centralizer, I mean the following, taken from the British North America Act, 1867. McDonald both had a guiding hand in writing it and implementing it.

1) Reservation and Disallowance. The Lieutentant Governor of a province was and is explicitly the agent of the Federal Government. The LG could reserve a provincial bill to the Federal Cabinet for assent or veto, or the Federal Cabinet could veto a provincial bill on its own through the LG. McDonald used this power frequently.
2) The Criminal Law was made a federal jurisdiction.
3) McDonald made sure that all superior court, county court and appellate court judges are federally appointed and paid. It's often thought that he did this to give the government a ready source of patronage in every riding.
4) Section 94: The federal government may pass legislation to make provincial property and other civil law including procedure uniform in the common-law provinces. This required the assent of the provincial legislature to the bill.
5) Judges explicitly may be appointed from any common-law province to any common-law province when the provincial civil laws are made uniform, following from above.

Determinant - the respect is appreciated and I hope to be able to show no less. However, I still don't think that your point is sufficient to throw away my suggestion.

For one thing, legacies ex-post are less relevant to my view than general expectations prior to elections, as I'm not attempting to defend a causal link from actual policies to economic growth, but only an interpretation of what I think is due to a limited but real coincidence between the median expectations of the voters in both economic and very general ideological respects. It seems to me that you would have to appeal to evidence that voters at the time expected less spending from a Liberal government than from a Conservative one, in order to hold McDonald against my hypothesis. And I'm not sure you've made that case. A Liberal free trade with enough regulatory twists could certainly be seen as more costly to the taxpayer than a Conservative infrastructure program, couldn't it? And what would the BNA look like if Liberals had been in power?

Furthermore, another problem with your criticism, in my view, is that you choose a specific counter-example in order to deny the relevance of a correlation over a pretty long period of time, and a correlation which also seems to fit some data from the US. Socialists in France at the time of Reagan-Thatcher-Mulroney could be another great counter-example, in my view. But it doesn't really help deciding whether it's the ceteris paribus hypothesis itself that is bogus or if we just need to specify some other variable that may explain away the counter-example.

Be sure that I defer to your historical expertise; I'm just not sure what sort of alternative explanation for the observed correlation it can support.

Determinant you are reading history through some tortured provincial rights lens. Who cares what the initial BNA had to say what did the privy council rule time and again? Until Canada gets a supreme court (not superior) who cars what the superior court said unless the the privy council took direction from it?

The study is bogus but your side narrative adds nothing as to why.


Travis: "The study is bogus " - read it before you condemn it, and give a reason for your condemnation.

Any further uninformed comments will be deleted.

You find the same thing in the US, especially in the stock market. See:


"As the election date approaches, there's a positive probability the next President will be Republican. The EPV of profits is rising as the election date gets closer. So stock prices are rising. And if we have a Republican president, there's a positive probability the next president will be a Democrat, and the EPV of profits is falling as the election date gets closer. If stock prices = EPV of profits, they will be rising under a Democrat president and falling under a Republican." -- Nick Rowe

That's been looked at, and it's not it. The key paper is from UCLA financial economists Pedro Santa-Clara and Rossen Valkanov, published in arguably the most prestigious journal in academic finance, the Journal of Finance, 2003, "The Presidential Puzzle: Political Cycles and the Stock Market". From the abstract:

The excess return in the stock market is higher under Democratic than Republican presidencies: 9 percent for the value-weighted and 16 percent for the equal-weighted portfolio. The difference comes from higher real stock returns and lower real interest rates, is statistically significant, and is robust in subsamples. The difference in returns is not explained by business-cycle variables related to expected returns, and is not concentrated around election dates. There is no difference in the riskiness of the stock market across presidencies that could justify a risk premium.

I don't accept that the data set is too small to be meaningful. First of all, 16 is definitely enough observations to make general statements of correlation, even if you can't reliably estimate the size of it.

Second of all this pattern appears in almost every other industrialized economy you look at -- it's not just the 16 you see in Canada.

It's not a coincidence at all. The question is why...

Richard - I just had a quick look at the paper you refer to, and it is quite fascinating. I'm not too sure what to make of the fact that they can't replicate the presidential effect with party dominance in Congress. But although I'm entirely incompetent in either macro or finance, I think I understood their main concluding conjecture, that Democrat presidencies systematically surprise investors for the better. And unless I completely missed their point, I think it's consistent with the story I'm inkling towards about what might be going on.

If both investors and voters have similar noisy info on the business cycle, and both groups also believe that one political label is more likely to have a significant re-distributive impact from higher incomes to the median taxpayer in good times, and from the median to the low income in bad times, it makes sense that a) the median voter votes Democrat only when he anticipates an upturn in the real economy; and b) the median investor fears this very phenomenon as much ex-ante, which means his expectations are lower than would otherwise be, but then the realized re-distribution is much less than anticipated, so that *unexpected* returns are higher under a Democrat president. Could that make sense?

This would mean, I think, that the solution of the puzzle would lie in the political rhetoric being quite convincing, to the point of renewing a persistent bias, election after election, in expectations as to the actual level of redistribution that will be undertaken. Neuro-economics anyone?

Yvan: "I'm not too sure what to make of the fact that they can't replicate the presidential effect with party dominance in Congress."

Earlier the contrarian federal/provincial tendencies were noted - Liberal federal gov'ts are often associated with conservative governments at the provincial level. And in the US one can argue that the Clinton's best years, for example, occurred when he was dealing with a republican congress. Interesting.

Frances - I would have thought the parallel with federal/provincial differences would be with federal/state policies in the US. The Congress vs President doesn't exist in our parliamentary system - this would mean the correlation is with the executive branch but not with the legislative. As to contrarian tendencies at any level, if there was something systematic about it, we'd still see these other levels being associated with our variables of interest, but with the opposite sign, right? And here there is no such association. So maybe it could point to what voters see as the main locus of power (federal governments, executive branch) versus institutions that they see merely as mitigating or reinforcing forces, which would kind of cancel out on average? Just a thought.


Again in the spirit of respectful debate, you're falling for Frances assumption that party positions are static. They are actually very dynamic.

Furthermore the concept of government redistribution of wealth wasn't part of the conversation in the 19th Century except for Marx. In Canada the question was whether to increase democratic government, including Responsible Government and a universal franchise. That question then morphed into development, a goal that was accomplished with railways and tariffs. In the 19th Century the Republicans, the British Tories and the Canadian Tories all supported high tariffs to favour domestic agriculture and industry. Remember at this time the Republican power base was the Northeast. Liberals and Democrats in all three countries were in favour of free trade (no protective tariffs, just low tariffs for revenue). This favoured small tradesmen and basic industries like Lumber.

In Canada there was also a very strong language/religion bias where the Tories had a power base in Protestant English Canada. Quebec was up for grabs, and first went Tory, then Liberal. Nothing to do with economics.

The political spectrum started to turn in the early 20th Century when the British Liberals tabled the "People's Budget" in 1911 and began the Welfare State in the UK.

Conservatives/Republicans were all for protective high tariffs until the Great Depression when that failed as a strategy for both economic prosperity. The advent of the Income Tax in all three countries is what enabled the Conservatives/Republicans and Democrats/Liberals/Labour to assume their present ideological stances. This was only reinforced with the advent of the Welfare State, a concept made possible by the Income Tax.

It's also good to remember that the Canadian Liberals are survivors. The British Liberal Party was reduced to third-party status by Labour in the 1920's. They only returned to power with the present Tory/Lib Dem coalition in the UK. The Canadian Liberals were scared stiff of the same fate and adapted by adopting the Welfare State as their own.

Your thesis assumes conditions that were simply not present in the 19th century and were in flux in the period between WWI and WWII.

It may hold after WWII but that says that conditions and political discussion were different before WWII. They were.

Sometimes a correlation is just a random coincidence. I say it's just a number, it has no foundation in history and no basis in fact or policy. History first, then the number.

Determinant - I'm not even sure we're having a debate here. It seems to me we're just moving in completely different spaces.

First, I really have no clue why you think I'm assuming as much about party positions. In fact, I could say that all I'm assuming is that the words Liberal and Conservative themselves have a semantic link, in the mind of the median non-scientific voter, with the parties' respective average spending propensities. The label itself could well be the simplest, most obvious element of a rhetoric which is supposed to make both parties appear as a real alternative to the other. That's really all I'm saying. What the real actions ex-post will be is not really relevant though, except if you could make the case that some other part of the rhetoric that had a lot more traction on votes implied that Liberals were perceived, as counter-intuitive as it may seem, to be *more* conservative with government resource consumption than Conservatives themselves. Note again that what counts for my "thesis" is what people expect, not what actually happens.

Second, as I think I've said before, every dollar a government spends is a dollar re-distributed. Otherwise there would be no need for government activities in the first place. And if no-redistribution is to be taking place, then pubic servants will have no paychecks. And as far as I know, the tax collector has not historically been viewed in the best of light, in any society. This is all I'm assuming here again, along with some self-interested voting. When people see an advantage in such redistribution (aka level of government activity), they tolerate more of it.

Third, on methods, correlation and things having no economic content. Well, what is at stake here is the part of the voting decision that has to do with economics, and whether this has traction on the vote or not. Whether there are unrelated determinants of that decision is... unrelated, right? So from the perspective of a correlation with an economic variable, all these differences between the parties are simply orthogonal/irrelevant/random. We're not interested in what is unrelated but in what is related. Random coincidences happen, sure, but when the coincidence is compatible with a coherent theory, I'd say the onus is on the person claiming this is an accident to suggest an explanation of why this accident could happen. For example, what sort of coincidence between non-economic policies, actual or expected on the one hand, and economic cycles on the other, could be at the source of this correlation, or else which outliers are driving this result, which actually appears in multiple countries over different time spans?

Finally, my own reference to the post WWII Welfare state only had to do with the fact that the justifying rhetoric for re-distribution from then on took the specific form of explicit income redistribution through monetary transfers to the poor or unemployed. But that has nothing to do with my attempt at explaining the observed correlation, so I don't know why we should further re-hash it.

This being said, I don't want to give the impression that history is irrelevant in my view, but it's not what is special about each individual event that is at stake when a correlation presents itself between them, it's the common ground that we are attempting to understand. The historian may see many distinguishing features from one event to another, but surely he/she must also need some structural assumptions, trends, general features of some kind in order to give some background to these singular things. And this is what we should be looking for, right?

Can a slip ever be so Freudian? You know what kind of servant...

Yvan, am enjoying your comments on this thread.

My pleasure. Maybe I'd have done nicely in an economics department, that's if I had not flunked my macro core at Western in the summer of 95. So I'll never know. But hey, I don't have to handle all them grant proposal red tape sour grapes, do I...

I think growth is simply more important to liberals who see it bringing prosperity, while preserving the status quo is most important to conservatives who view change as loss. Their most important issues are not economic but social and reactionary, and preserving the social order.

In postwar Britain the story runs the other way round - it is the Tories who typically see higher growth (and lack of data spares Labour 2009 and 2010). Lets look at per capita postwar British growth in real terms(I'll give the election year to the outgoing party)...

Atlee (Labour, 1945-1951): 1.1%
Churchill/Eden/MacMillan/Home (Tory, 1951-1964): 2.5%
Wilson (Labour ,1964-1970): 2%
Heath (Tory, 1970-1974): 2.7%
Wilson/Callahan (Labour, 1974-1979): 2.7%
Thatcher/Major (Tory, 1979-1997): 2.1%
Blair/Brown (Labour, 1997-2008): 1.9%

In Australia as well, the right-leaning Liberal party has the better record of per capita growth since 1960
Menzies/others (Liberal/Country, 1960-1972): 2.5%
Whitlam (Labour, 1972-1975): 0.44%
Fraser (Liberal, 1975-1983): 0.92%
Hawke/Keating (Labour, 1983-1993): 1.4%
Howard (Liberal, 1993-2007): 2.2%

(and Loss, that view does not square well with the past 30 years, which often featured the left fighting to preserve the welfare state, while radical reformers were on the right).

A-ha! Now that is a challenge! Many thoughts here: make sure we're looking at the data the same way in the different countries, make sure we control for similar factors - for example, the Canadian study controls for US growth i.e. main trading partner or proxy for something else? - and eventually, maybe look at some specific items of economic rhetoric pre-election and/or policy post-election to see if something else than the conventional right-left split could be driving all of this.

One thing that comes to my mind is how North American main parties may not have been as closely associated with the working class, as opposed to the capitalist elite, in the way Labour parties have been? That would be more like having the NDP in power here and Ralph Nader as president south of the border, right? So maybe causality could run in the other direction after all? Governments may have no clue what is good policy, yet some parties may have good recipes for disaster?

Another possibility, so that my preferred hypothesis may remain relevant, could be related to higher rates of unionization in the UK and Australia at least prior to the 90s. The more unions are seen as the mechanism of choice for re-distributing wealth, the less likely that politics may influence the relevant expectations. This still can't account for the reversal though (that is if such reversal remained once controlling for other relevant factors). And what if rich capitalists were in fact seen as big bad wolves there by the median voter? Could he/she feel that a Tory government could only be trusted in good times - otherwise the transfers may very well go from the middle of the income distribution to the top?

Sure keeps the whole matter completely wide open.

In Canada the Liberal vote is more likely to be female, urban, young. In Britain I don't think the Labour party has nearly such a strong lock on the female vote as, say, the Democrats do in the US, and also there's lots of older Labour supporters. So the British experience to some extent fits underlying demographic/interests type stories.

On Australia: clearly more research is needed - perhaps a walking tour of the south Australian coast to really assess the situation...

But Frances - how do demographics/interests explain changes - or more to the point how can long-term change in demographic variables affect short-term voting and economic patterns?

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad