Saturday, December 12, 2009

Another PR Exercise

Back in July, I did a proportion representation exercise where I portioned out seats according to the 2008 election results and the current make-up of parliament. In other words, that PR exercise assumed 308 seats with the same provincial representations as they currently have in parliament.

In this exercise, I took it a step further. I divvied up the House of Commons into 332 seats, or one seat for every 100,000 Canadians. I then divided up the provinces according to that principle.

When doing so, I got 129 seats for Ontario (38.9% compared to 34.4% currently), 78 for Quebec (23.5% compared to 24.4% currently), 44 for British Columbia (13.3% compared to 11.7% currently), 35 for Alberta (10.5% compared to 9.1% currently), 12 for Manitoba (3.6% compared to 4.5% currently), 10 for Saskatchewan (3.0% compared to 4.5% currently), 9 for Nova Scotia, 8 for New Brunswick, 5 for Newfoundland & Labrador, and one each for Prince Edward Island and the territories (all three together). So, Atlantic Canada would have 6.9% of the seats in the House of Commons as opposed to the 10.4% they currently have.

Breaking it down this way, we see that the most under-represented provinces in Parliament today are Ontario (who should have around 120 seats of the 308), British Columbia (who should have 41 instead of 36), and Alberta (who should have 32 rather than 28). Contrary to popular opinion, Quebec is actually not significantly over-represented in the House of Commons. Only Atlantic Canada, who should have 22 seats in the HoC rather than 32, are punching well above their weight.

Using these figures, and portioning out seats according to the 2008 electoral results (always rounding as necessary, and giving the higher ranked parties priority) we get the following result:Obviously, still a Conservative minority. But rather than 46.4% of the House of Commons being occupied by the Conservatives, only 38.9% of it is.

In this scenario, the Liberals and New Democrats, with 150 seats combined, could out-vote the Tories. But they would still be short of a majority, and would require the support of the Greens or the Bloc Quebecois (significantly reduced, as you can see) to pass legislation.

Conservative support is mostly divided between the West (55 seats) and Ontario (51). They also have a good-size caucus from Quebec (17) and a few Atlantic Canadian MPs (6).

The Liberals have their power-base in Ontario, with more than half of their MPs from that province. Quebec and the West each have almost equal representation, while the Liberals would have the most Atlantic Canadian MPs.

The NDP would be, like the Conservatives, evenly divided between the West and Ontario (22 and 24 MPs, respectively). They'd also have 10 Quebec MPs and 7 in Atlantic Canada.

The Bloc is the big loser in this new configuration, as they would be reduced to only 30 seats.

The Greens are the big beneficiaries here, going from 0 seats to 23. They are also evenly divided between the West (9) and Ontario (10), but have three Quebec MPs and one from Nova Scotia.

Anyway, I find these alternate scenarios fascinating. What a difference it would make in Canadian politics!


  1. "What a difference it would make in Canadian politics!"

    I wish somebody would do a study, or perhaps they have and I haven't seen it, about what effect these sort of governments have on debt levels.

    Ira and myself have both hypothesized that the constant buying off of smaller parties to form a majority and the sudden importance of individual members might create a sitution where money is used as a sort of bribe.

    You might also see a situation where parties refuse to make tough cuts because there's too many opponents who could demagogue you.

    Would it turn us into Greece, Italy, or Japan ?

    I don't know, but its something worth discussing!

  2. When BC was having their first referendum on electoral reform (why we had the second referendum I have no idea), there were those sorts of studies being published.

    I think The Fraser Institute did one - I know I saw a presentation by someone there arguing that proportional representation has typically produced bigger government.

  3. Pure PR would make a huge difference, but I don't think we'd ever see this in Canada. What would be interesting is looking at the 2 systems that have been voted on in recent years, MMP and STV. I know there's probably not enough data to do an STV analysis, but we could certainly look at MMP. The somewhat obvious outcome would be somewhere between what we actually had and this PR outcome you just calculated.

  4. Japan is a poor example because it has been run almost totally consistently by majorities despite it's proportion based Mixed Member system.

    A comparison is actually very difficult, because among advanced economies there are only a handful of first past the post systems -- the US, Canada, and the UK. There are also runoff/transferable systems in France and Australia.

    Using the IMF table for 2009, (below) we can see that all of these states are below the average for advanced economy G-20 for the years being recorded or projected.

    Among the emerging economy G-20, only Mexico has a FPTP system. (I'm not going to bother with Saudi Arabia, China, and Russia for reason I shouldn't to explain) Mexico seems above the average for the emerging G-20, but it's actually below some of its contemporaries. The autocrats kinda throw the average.

    However; we can not only see huge variations among the plurality systems in terms of Debt/GDP ratio, but we can also see many party-list and MMP system doing much better or the same as our system. Including Germany, Spain, South Korea.

    Bigger government would result in Canada under PR because of the greater amount of left wing parties... but bigger government and debt levels don't necessarily go hand in hand.

    Convenient Link to table:

  5. Kevin,

    a direct comparison between countries with and without PR seems useless given the individual variability between countries.

    What might be more usefull is looking at a specific country that made a switch between systems and compare the average growth in debt from the 20 years previous to the 20 years after the switch compared to the average growth of debt in all similiar countries.

    I don't know how many situations there are like that. There probably isn't enough data to make that kind of straightfoward analysis.

    Any sort of study comparing countries would have to be very, very complex and control for all sorts of factors - including the success of left wing vs right wing parties.

    A very right wing country might not see a rise in spending because parties would try to out compete each other as the most fiscally responsible.

  6. I'd be interested in seeing a hypothetical concerning the use of a STV-like, or maybe IRV, system. Maybe like the Australian system, where party support is more or less looked through the glass of competition between Labour and Liberal?

    It is possible to do something like that as well, as there is a multitude of polls out which ask for a person's second-party preference.

  7. This comment has been removed by the author.

  8. This is hardly consequential, and open to lots of potential methodological criticisms, but, with those caveats aside:

    I'm working on a paper (just undergrad term paper stuff) and had this data at hand, so, here is a basic linear regression looking at the issue:

    lm(formula = Cengov2004 ~ democpoplog + fedtype + GDP2004 + Urban2004 +
    Elecpr + Elecmaj)

    Min 1Q Median 3Q Max
    -9.834 -2.894 0.112 3.140 11.821

    Estimate Std. Error t value Pr(>|t|)
    (Intercept) 3.090e+01 6.620e+00 4.668 1.68e-05 ***
    democpoplog -1.105e+00 3.694e-01 -2.992 0.00398 **
    fedtype -1.239e+00 1.499e+00 -0.826 0.41194
    GDP2004 9.335e-05 5.832e-05 1.601 0.11456
    Urban2004 4.829e-02 3.834e-02 1.260 0.21253
    Elecpr 1.464e+00 1.637e+00 0.894 0.37458
    Elecmaj 2.824e-01 1.913e+00 0.148 0.88311
    Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

    Residual standard error: 4.83 on 62 degrees of freedom
    (20 observations deleted due to missingness)
    Multiple R-squared: 0.2442, Adjusted R-squared: 0.1711
    F-statistic: 3.34 on 6 and 62 DF, p-value: 0.006477

    So, controlling for population, GDP, Urbanization, and a dummy variable for a federal system, a dummy for PR and a dummy for Majoritarian systems both show small and statistically insignificant effects on central government spending (as a percentage of GDP; the data was only available for 60 of the countries in the data set unfortunately), I checked a similar variable for degree of proportionality, but it is less complete; however, it still shows no significance and a small coefficient. Ultimately the only significant variables in this model was population size. I know the formatting gets a little messed up on the comment board, but first column after the variable names is the coefficient, and stars indicate significance levels. I also find it somewhat interesting that federalism doesn't have a significant effect, even on its own it doesn't show one (Pearson R: 0.064, p-value:0.5998).

  9. You can't compare different countries to determine whether one type of government selection mechanic leads to different governance outcomes. There would be no way to correct for cultural biases.

    Even within Canada this should be obvious. If you were to compare Alberta to Quebec, you would see huge differences in how much interference there is in private enterprise, and how much regulation there is, and how much funding cultural endeavours receive. And yet they have the same electoral system.

    To determine if there's an effect caused by the electoral system, you'd need to examine the size of government over time in coutries that changed electoral systems. Only then would you be isolating the electoral effects.

  10. Hey Colin,

    interesting stuff. I have a fairly major quibble though.

    You're measuring the effect of PR on government spending ? Without taking into account revenue?

    Is there any way you could measure the effect on total spending compared to total tax revenue (new debt in real dollars). And then calculate the effect on the growth of the debt to GDP ratio?

    I think this is nessecary because it seems to me that certain countries have a preference for large spending on social programs or military spending that has nothing to do with the form of gov't.

    Considering the willingness to go into debt, as opposed to the willingness to tax and spend, seems like a realistic way to control for the political leanings of a country, a key variable when considering the size of spending.

  11. Hey Jesse,

    Thanks for your reply. Unfortunately the variable for tax revenue is woefully incomplete in the dataset I'm using so I can't really look at the difference between revenue and spending in any significant way. I do think that because the variable is calculated as a percentage of GDP it does show the size of the government as a percentage of the GDP, and that this is an interesting thing to examine also. What I mean is that there are two slightly different things that we are talking about measuring. You seem to be interested in the relationship between electoral systems and fiscal responsibility, I was just measuring the relationship between electoral systems and total size of government. Both are legitimate, but I think separate, concerns.

    Also, in general, I think that even though this particular model is just cursory and not at all meant to be conclusive or robust, I find nothing wrong in principle with using this sort of cross-national model to measure the effects of electoral systems. Surely cultural differences are likely related to spending, however it doesn't strike me as likely that they are related to electoral systems in such a way as to confound this model. Surely a proper model would control for them, and no model can ever claim to account for all the variation in the dependent variable, but I think that if we are hypothesizing that electoral systems likely have an effect on spending, fiscal responsibility, debt, or whatever; than that effect should be observable between different countries if we control for the relevant factors.

  12. Colin,

    I have nothing against cross country comparisons provided you have a model complex enough, that controls for as many variables as possible.

    My thinking on this is that PR in and of itself most likely wouldn't alter spending in a manner large enough to be statistically significant.

    It would, of course, increase spending significantly in Canada but that's only because we would elect less Conservatives and give greens seats.

    But in somewhere like Alberta it might give a party like Wildrose a big chunk of seats and tilt the system rightward so as to reduce spending.

    The method by which PR would increase spending would be through earmarks/pet projects/pork. A fractured parliement would require cash for various ridings to bring together a majority on a budget.

    This type of spending would occur around the edges of a budget and would tend to be very small in size, I think earmarks are like 1% of the American budget and they are rife with them.

    The other thing is that this spending could be HIDDEN in inefficiency - the total spending stays the same but money is directed at extremely sub-optimal projects to satisfy politicians.

    Perhaps an easier way to test this is to see whether majority parliements tends to spend less or more than minority parliements ?

    The tendency to elect minorities is the only mechanism by which I would see PR increasing spending in and of itself.

  13. Eric would it be possible to do this PR by adding the seats provincially and then using the first past the post method to determine the seat totals. That's the information I'd like to see. PR seems to me to be begging for indecision on a European style. I don't want the pizza parliaments of Italy nor the tail wagging the dog parliaments of Israel. I want governments that can and will make tough choices even if I disgaree with them. Governments that make bad choices normally suffer defeat at the ballot box the next time out. That's as it should be.

  14. There isn't much doubt that we are in a period of GW. I question how much man has to do with it however. There have been periods in earth's history that have been warmer than anything we may experience now. There are numerous theories as to what is responsible for GW. There I think the science is still out.

    As for the cap and trade system being proposed this seems like a thinly disguised method to transfer wealth from the developed world to undeveloped world.

    Dion's carbon tax was the most direct and effective way to try and influence people's carbon decisions. If it had been a strict transfer of carbon tax dollars to income tax cuts it might have gained some traction. That's if you believe taxing carbon will any way be effective in slowing GW. Much of what I've read says we're too late to effect change and our money would be better spent trying to figure out how to adapt to GW.

    The one issue in GW that everyone avoids is the biggest one IMO. That is the issue of population growth. Until we stop population growth we have very little chance to deal with GW. We are just like lemmings heading for the edge of the cliff. If we are unwilling to stop population growth, then nature will do it for us, through war over food, water and resources, disease or a toxic environment.

  15. Earl,

    The issue of too many minor parties might not be so bad if the seat totals were Canada wide instead of provincial. Thus the 9% BQ national vote would be 9% of the seats in parliement.

    (I'm just being cynical, i'm actually pretty die hard about first past the post. But it is a valid arugement - why abstract vote counts to the provincial level instead of some kind of regional total or Canada wide total.)

  16. Fascinating article in the globe and mail suggesting we should all adopt the one child policy China has.

    Too bad the article was written by a woman with two kids...

  17. Jesse,

    Sorry for any confusion, the second paragraph was meant to address some of the other comments left, not yours. I tend to agree with your analysis of the effects in Canada given the current voting patterns. However, while it is certainly likely that fractured legislatures can tend towards parochial, and thus pork-barrel, politics, I think a well designed system can attempt to address this while still reaping the benefits of PR. Looking at the US we see that there are other systemic factors that lead to gross overspending and pork-barreling, it being one of the most majoritarian systems in the world. PR systems with very large, or national lists on the other hand can tend to elect legislators with unduly national outlooks, forgoing the need for any local pork. I think that in Canada the level of party discipline, even if it were to decline from the high it is at now, would mean that any party involved in budgeting with national representation would have just as much incentive to seek pork as it currently does. In fact, as it is now, a regional party like the Bloc, which is more likely to have parochial interests, can have more influence than they arguably should and demand more than they would be able to in PR.

  18. Jesse said,
    "It would, of course, increase spending significantly in Canada but that's only because we would elect less Conservatives and give greens seats."

    Why would Greens in Parliament increase government spending? The Greens I know are all fiscal conservatives. Greens lean towards the libertarian end of the libertarian-authoritarian spectrum. These combine to suggest a balanced and stable or declining budget. In fact, the opposite of what we were seeing from the current government even before the recession.

    A concrete example: under Frank de Jong, the Green Party of Ontario came out against subsidies to companies generating renewable power. The party proposed levelling the playing field for feed-in generation, but beyond that, letting the market determine the price of power--by properly charging for all costs. If a company could make a profit with a nuclear plant after paying for proper waste management, good. If a company could make a profit from coal after sequestering or paying for the carbon generated, fine. And if wind or solar then made the most sense, we'd build wind or solar plants. Consumers wouldn't pay any more for power--they'd just see it on their bill instead of hidden in their taxes as subsidies or cleanup costs. And they'd have the option of paying less by using less. The market would send a clear signal.

    The Green Party is not left-wing.

  19. John,

    A quick look at the website of May's Green party turned up plans to massively increase social spending:

    "The Green Party of Canada is committed to ... a national affordable child care program, programs to eliminate the financial barriers to post-secondary education, programs for affordable housing... is time to revisit a major policy initiative -- the use of a negative income tax, or guaranteed liveable income (GLI) for all."

    Worse, of course, is that at the same time we'd see a sharp reduction in revenue as burdensome environmental regulations are introduced and the oil sands are shut down.

    Not only this but she is calling for tax cuts and income splitting!

    Talk about pie in the sky. These policy initiatives would result in a disturbingly unbalanced budget.

  20. GW: There isn't much doubt that we WERE in a period of global warming from 1940 to 1998. We have good instrumental data from that period, and aside from a slight drop in the 1970s (which caused the whole global cooling scare) there was a steady increase over that whole period.

    But global temperatures have yet to exceed those measured in 1998. 1998 was the warmest year on record (records that go back only to 1940), and it still is. We've had 11 years of no warming.

    We also don't know if those temperatures we're seeing are unusual. All the temperature numbers before 1940 are proxy data assembled using dendrochronology (tree rings), but as the leaked e-mails point out (and Steve McIntyre had already demonstrated this) the proxy data shows a sharp DECLINE in global temperatures after 1960. Since we know that global temperatures haven't been declining over that period, this tells us that dendrochronology is a lousy proxy for global temperatures. As such, we have NO IDEA what the Earth's temperature has done prior to 1940. There may well be good ways to find that out, but the climate researchers who are feeding data to the UN aren't using them.

  21. Ira you are wrong on both counts. The common myth that there hasn't been any warming on the last 11 years is false and based on the fact that just the US hasn't seen warming even though globablly it has been a huge problem, the last few years have all been among the 5-10 warmest years on record. Further, it is absolutely not true that dendrochronology is the only proxy for historical temperatures, ice core samples and melting rates, among others, are all used (by NASA for example) and all corroborate both the historical dendrochronology data and the general findings about the trends of current warming. Further, there are recorded temperatures prior to 1960 and the dendrochronology matches up with those also. The decline after 1960 in dendrochronology readings has been widely known about and discussed in the literature, and while by my understanding it remains something of a mystery (I think there are some vague environmental change theories) that doesn't change the fact that there ARE other methods to measure historical temperatures, and they all show the same thing.

  22. Colin,

    Ira is completely right about 1998. The US doesn't record data for the continent alone. They have polar research stations and recieve data from around the world.

    While it may be true that some parts of the world have had trouble with warm temperatures that is WEATHER and not a measure of the global climate, which is by definition an average.

    On the second point, you are right to bring up ice cores.

    Unfortunately for you the ice seems to back up Ira.

    The Greenland sample shows a warming period that has more or less leveled off.

    It also shows periodic warming and cooling periods over the last thousands of years.

    The increase in temperature is unremarkable. In fact we're not even as warm as the medieval warming period yet.

  23. Not to mention distant history.

    The current average global temperature is about 14.5°C.

    During the Jurassic period, the global average temperature was about 22°C.

    And yet, life thrived. There was tremendous biodiversity. If our goal is biodiversity, then we should want a warmer Earth. Ice is the enemy of life.

  24. Colin is correct there the warmest year on record for North America differs from the warmest year on record for the world.

    But the warmest year for North America was 1934, thus demonstrating that local variability considerably overshadows any global trends.


COMMENT MODERATION POLICY - Please be respectful when commenting. If choosing to remain anonymous, please sign your comment with some sort of pseudonym to avoid confusion. Please do not use any derogatory terms for fellow commenters, parties, or politicians. Inflammatory and overly partisan comments will not be posted. PLEASE KEEP DISCUSSION ON TOPIC.