Effect of fuel prices and unemployment rates on road deaths

antoobrien

Iwannahurl wrote: »

Incorrect. See OP and post #5.

The text I quoted is taken directly from the link provided by the O.P.: https://crawford.anu.edu.au/acde/publications/publish/papers/wp2014/wp_econ_2014_18.pdf

To further reiterate, I went to the ANU library to search for this paper as a peer reviewed, scholarly paper - it's not listed.

They acknowledge that gasoline price is one determinant among many, and they attempt to model or predict the size of the effect by controlling for other variables. They then attempt to quantify the potential impact of a policy-based change in that modifiable variable.

The authors are suggesting policy based on a single factor - fuel price

They ignore the fact already pointed out that at the same price range there are significant differences in fatality rates.

They ignore studies that state that higher fuel prices could contribute to a higher fatality rate (haven't read them, they are Prima facie as plausible as the proposal)

There are ways in which higher gasoline prices might actually lead to more rather than fewer road deaths. One is: by reducing congestion, higher gasoline prices can allow remaining drivers to travel at faster speeds (Burger and Kaffine 2009), which increases the risk of fatal crashes. Substitution to particularly risky types of fuel-efficient vehicles, such as motorcycles, may also cause additional road deaths when gasoline prices rise (Hyatt et al. 2009; Wilson et al. 2009).2

They appear to ignore the impact of infrastructure in their approach (section II) - there is no mention of it, while they do call out terrain as being an issue

They account for infrastructure in the results (III) by suggesting it enables road spending therefore deaths (this would definitely be counter intuitive to the Irish case where spending on roads woulds seem to have been a factor in reducing the fatality rate) - without accounting for the quality of that infrastructure.

Here is an excerpt of their conclusion:

The international community is mobilizing a number of strategies to improve road safety during the United Nations’ Decade of Action for Road Safety 2011–2020. The Plan for the Decade of Action (WHO 2010) is silent on the potential role of fuel pricing

Can you see why one might get the impression of a predetermined result?

L1011

Iwannahurl wrote: »

However, for those of us outside the process the way to go about it is to demolish the actual study, not to make glib assumptions about the authors and reviewers.

So why did you attempt to shut down debate by making reference to peer review and publication and implication that those outside can't realistically challenge due to that?

Once again - you can't have it both ways.

Iwannahurl

L1011 wrote: »

So why did you attempt to shut down debate by making reference to peer review and publication and implication that those outside can't realistically challenge due to that?

Once again - you can't have it both ways.

I referred to such matters because of the likes of this:

Tragedy wrote: »

Did you even read that report?

I'm guessing not.

It is laughably statistically unsound at the most basic of levels. It didn't isolate for any factors other than fuel price. It didn't analyse the effects on internal changes in fuel prices on road deaths ceteris paribus, it admitted itself that few other studies held oil price changes to be significant or relevant when talking about road deaths.

All it did was take 144 countries, get the price for fuel, get the annual road deaths and compare them. That's it.

I seen you're similar to Iwannahurl in that you like to throw around facts and papers that you haven't researched or even read.

I didn't try to "shut down debate". Quite the contrary, which is why I'm still waiting for an answer to this post.

Iwannahurl

antoobrien wrote: »

The text I quoted is taken directly from the link provided by the O.P.: https://crawford.anu.edu.au/acde/publications/publish/papers/wp2014/wp_econ_2014_18.pdf

To further reiterate, I went to the ANU library to search for this paper as a peer reviewed, scholarly paper - it's not listed.

It's the same study as far as I can see: http://www.readcube.com/articles/10.1111%2Fecin.12171

If you're trying to suggest that their research was not published in a reputable journal, then you are wrong.

antoobrien wrote: »

The authors are suggesting policy based on a single factor - fuel price

No, they are focusing on a single factor which can be studied in order to provide evidence to support one aspect of public policy. That does not preclude other research or other policies.

antoobrien wrote: »

They ignore the fact already pointed out that at the same price range there are significant differences in fatality rates.

They ignore studies that state that higher fuel prices could contribute to a higher fatality rate (haven't read them, they are Prima facie as plausible as the proposal)

They appear to ignore the impact of infrastructure in their approach (section II) - there is no mention of it, while they do call out terrain as being an issue

They account for infrastructure in the results (III) by suggesting it enables road spending therefore deaths (this would definitely be counter intuitive to the Irish case where spending on roads woulds seem to have been a factor in reducing the fatality rate) - without accounting for the quality of that infrastructure.

Where in the study are they ignoring "the fact that at the same price range there are significant differences in fatality rates"?

Which studies "state that higher fuel prices could contribute to a higher fatality rate"?

Road infrastructure is not ignored, and is controlled for in their study along with their use of an Instrumental Variable (IV) approach. This level of statistical analysis is well beyond my (imaginary) pay grade, but my understanding is that their strategy is to demonstrate the relationship between gasoline price and road deaths first, and then to control for other variables that could explain the "substantial variation in road fatality rates among countries with similar gasoline prices".

antoobrien wrote: »

Here is an excerpt of their conclusion:

The international community is mobilizing a number of strategies to improve road safety during the United Nations’ Decade of Action for Road Safety 2011–2020. The Plan for the Decade of Action (WHO 2010) is silent on the potential role of fuel pricing.

Can you see why one might get the impression of a predetermined result?

No.

Here's an excerpt from their introduction:

Road safety is a leading public health issue. Road crashes are the cause of 1.3 million deaths every year; the ninth-leading cause of death globally and the number-one cause of death for people between 15 and 29 years of age. Road death rates are particularly high in middle- and low-income countries, which each year see an average of 20 and 18 deaths per 100,000 population, respectively. There are around 9 road deaths per 100,000 population each year in high-income countries. Up to 50 million people worldwide also suffer non-fatal injuries each year, bringing large human and financial costs. The global road death toll is expected to increase to around 2.4 million per year by 2030 in a business-as-usual scenario, making road crashes the fifth-leading cause of death. Finding ways to reduce global road deaths is an increasingly important policy imperative.

There's nothing wrong with using Economic theory to inform potential policy responses to such a serious Public Health issue. It's science in the public interest, imo.

galwaycyclist

The international community is mobilizing a number of strategies to improve road safety during the United Nations’ Decade of Action for Road Safety 2011–2020. The Plan for the Decade of Action (WHO 2010) is silent on the potential role of fuel pricing.

Too busy with work at moment to contribute meaningfully. But this triggered a side observation.

One of the World "Health" Organization's primary partners in their "road safety" initiatives is the FIA Foundation.

http://en.wikipedia.org/wiki/FIA_Foundation_for_the_Automobile_and_Society

Who are the charitable arm of world motor sport.

Its a bit like having producers of alchoholic drinks as a main partner for anti-alchohol initiatives.

antoobrien

Iwannahurl wrote: »

It's the same study as far as I can see: http://www.readcube.com/articles/10.1111%2Fecin.12171

If you're trying to suggest that their research was not published in a reputable journal, then you are wrong.

I'm not the university search engine is. Since they have multiple papers cited from both authors and this isn't listed.....

Iwannahurl wrote: »

No, they are focusing on a single factor which can be studied in order to provide evidence to support one aspect of public policy.

Yeah, see the problem is that when critically analysed there are so many holes in their "evidence" that their policy can't actually be supported.

Iwannahurl wrote: »

Where in the study are they ignoring "the fact that at the same price range there are significant differences in fatality rates"?

Which studies "state that higher fuel prices could contribute to a higher fatality rate"?

Did you read either the article or the posts you're querying, because it doesn't look like it.

Iwannahurl wrote: »

Road infrastructure is not ignored, and is controlled for in their study along with their use of an Instrumental Variable (IV) approach.

I suggest you read the report again, their results ignore the quality of the infrastructure in favor of "Road distance".

Iwannahurl wrote: »

There's nothing wrong with using Economic theory to inform potential policy responses to such a serious Public Health issue.

If the research is valid, then yes. However this is not anywhere approaching valid.

galwaycyclist

Little discussion doc from Gerald Wilde from 1989 I think

http://www.icadtsinternational.com/files/documents/1989_010.pdf

I was actually trying to google Partykas work but this will do. The figures are at the very end of the document.

The influence of the economic juncture

Although these findings seem to be roughly in line with the alternative view depicted above, there is reason for caution in accepting them as convincing support. An unfortunate feature of before-after comparisons without external control data, and likewise of the inspection of time-series statistics, is the possible impact of other factors that are not accounted for. One such other factor that has recently begun to attract special attention is the fluctuation in the nation's economy. Its effect in the USA has been explored by Partyka (1984), in Great Britain by Adams (1985) and in British Columbia by Mercer (1966). These authors found economic recessions to coincide with reductions in accidents.

In order to add to the existing information on this matter, annual rates of unemployment were investigated on their correlation with the same-year traffic death rates per capita in seven different countries (data derived from the International Labour Office, various years, the US Department of Labor and national statistical yearbooks). These countries were selected because of relatively long time-series of unemployment statistics being available, relative consistency in the manner unemployment was defined and measured, large population size (reducing mere statistical chance fluctuation in the data) and - frankly - also because I was to address audiences on the topic in some of them.

Figure 3 relates to a 40-year period in the USA. Close scrutiny of the changes from one year to the next reveals that in most cases an increase in the unemployment rate is associated with a decrease in the per capita fatality rate in traffic. The top profile is virtually the inverse image of the bottom profile. The product-moment coefficient of correlation, which is admittedly a questionable measure of association in cases of long-term secular trends in one or more of the variables (in the present case not so much in the death rate but in the unemployment rate), amounts to r=-.68.

In the other six countries, as Table 2 shows, the correlations were found to be stronger. In Canada, the United Kingdom and the Netherlands they amounted to r— .88 . There are various reasons to suggest that the observed correlations may be underestimates of the true correlation between the unemployment rate and the per capita traffic accident rate. The unemployment rate is usually estimated from samples, household surveys or otherwise limited data bases. That estimate is thus subject to error. The same holds for the estimate of the number of people killed as a consequence of a road accident (Hutchinson, 1987). The number of people residing in a country in any given year is not known for sure either. The calculated coefficients of correlation are thus likely to suffer from attenuation due to unreliability of the data. Time lags between variations in the one and the other would also have an attenuating effect. On the other hand, the coefficient would be inflated if the errors in measurement in both variables are correlated (i.e., autocorrelation).

To complicate matters further, over the years there have been changes in various countries in the definition of what constitutes "being unemployed". In some countries, migrant labour is sent home when unemployment rises, thus changing the number of residents and its distribution in terms of age and social class (and thus accident likelihood).

The last part might be quite relevant to Ireland since we had a mobile migrant community and also a "native" workforce who react through emigration.

galwaycyclist

Tragedy wrote: »

I see, following in IWH's footsteps. Don't address criticism of previous report that you had posted and obviously not read, post another report that you haven't read instead to change the goalposts.

I could explain the parts of the report (that you haven't read) that are very obviously caveated by the authors, I could point out numerous leaps in the report that weren't warranted. Indeed, I could write out a long and detailed reply that you would ignore, but there's no need really is there?

You posted about a correlation between two statistics as if that showed something important. Any statistics student is taught in the very first lesson that Correlation does not equal Causation.

So, how about stick to what you're good at - which isn't reports, statistics or any kind of information beyond 'I hate motorists'? It would save us all another useless thread where IWH gets to do what IWH gets to do.

If we were talking about one very unusual report with an unusual finding then there might be merit in getting into the details of the method used.

However - the observation that changes in unemployment rates - used as a proxy for economic activity - are associated with inverse changes in road deaths is not new. It has been around for years and is reported by various respected researchers.

Of course the authors enter caveats about their findings it would be unacceptable not to. What is your point? Are you trying to allege that admitting the existence of confounding factors invalidates such analyses? If so, then in my view you are just being silly.

You are correct that correlation is not causation - but when the same correlation is seen across varying populations and time periods then it provides strong evidence of association whether direct or indirect.

Iwannahurl

antoobrien wrote: »

Yeah, see the problem is that when critically analysed there are so many holes in their "evidence" that their policy can't actually be supported.

If the research is valid, then yes. However this is not anywhere approaching valid.

Repetition is not a valid form of debate.

I think we can trust the authors' mastery of statistical analysis more than we can rely on yours.

Iwannahurl

galwaycyclist wrote: »

correlation is not causation - but when the same correlation is seen across varying populations and time periods then it provides strong evidence of association whether direct or indirect.

A particular study may not be designed to determine causation. Here's an Australian study by researchers at Monash University in Victoria, which explored factors impacting on the 29% decrease in road fatalities in 1990 compared to the previous year, as an economic downturn occurred.




The authors are explicitly stating that their study cannot establish a causal link, but their statistical model explains a sizeable proportion of the observed reduction in fatalities, and shows that a quarter of the decrease was due to economic factors.

In Epidemiology/Public Health there are criteria for causation, traditionally listed as follows:




The association between fuel price, unemployment and road fatalities is well established. In terms of a causal pathway, one plausible explanation is that being unwaged and/or having to pay a high price for fuel could lead to a number of situations that would reduce fatality risk on the roads:

• slower and more conservative driving, to save fuel
• purchase of smaller, and hence more fuel-efficient, vehicles which also have lower mass
• less driving overall
• less leisure or social driving, which may be associated with alcohol consumption

Prof Michael Sivak of the University of Michigan has researched these possible factors:

http://ns.umich.edu/new/releases/7221
http://ns.umich.edu/new/releases/7184
http://ur.umich.edu/0809/Jun22_09/08.php

Tragedy

Iwannahurl wrote: »

Is that sound I hear the moving of goalposts?

What you said earlier was:



That was clearly untrue. Now you're attempting to argue that, while other variables were indeed controlled for, they were neither correct nor useful.

My apologies. They failed to account for any factors relevant other than price of fuel. Infant mortality is not relevant. Length of unpaved road network is not relevant. Etc etc. I already listed some common factors that time and again are considered relevant to road facility statistical research, and none of them were used. Most likely because they would have significantly altered the studies findings and rendered it both meaningless and unworthy of publishing.

There is nothing in the WHO guide you linked concerning road length per 1000 population, so would you care to tell us why this variable is more correct and useful, and what difference it would make to the analysis?

For someone who likes to talk as if they understand statistics, the fact that you don't understand why absolute figuresbare not comparable across different populations is quite hilarious. Ireland has less absolute road deaths than England, let's ignore that England has a vastly higher population and assume that Ireland is safer!

As for your other comments, I know what peer review is

You have shown multiple times in this very thread that you don't understand peer review.

Please do not lie.

The basic point is that contributing to a chat forum such as Boards is not comparable to the process by which academic studies are assessed, published and critically reviewed.

That's akin to saying that Paul Krugman can never discuss a theory or economic policy in a newspaper, interview or debate because it is not comparable to the process of peer review.

[QuoteTherefore it's at least mildly amusing to see someone such as yourself, in a chat forum, pompously declaring that a published paper is "laughably statistically unsound at the most basic of levels". That is all.[/quote]

The only mildly amusing thing is that you have verifiably demonstrated that you don't understand ceteris paribus, that correlation does not equal causation, that you can't compare absolute statistics which is why we generally compare statistic per <unit> and that you don't understand that if x increases by 1% when y decreases by 1%, x decreases by 1% when y increases by 1%.

In short, you have verifiably demonstrated time and again that you understand very little of the arguments that you are making, and none of the reports you are parroting.

This is an empirical fact Iwannahurl.

Also, in any case, what is the relevance of the 2006 MIT paper to the 2014 study cited by the OP?

I already explained that. Please read posts before you reply.

You've misquoted the US EIA.

I misquoted no one. Please do not lie.

Why did you not provide a link so that your claim could be checked out?

I didn't make a claim. I reported a claim the EIA made. Another basic error IWH.

Why did you change it from decrease/raise to increase/reduce?

Because it means the same thing and it genuinely scares me that such a very basic lack of knowledge in how two related properties with an inverse effect on each other r work, makes you think you've won some sort of victory.

Why are you comparing the US EIA short term estimates with the "long-run gasoline price elasticity" estimates which are the subject of the study cited by the OP?

It is important to note that these results measure consumers’ reac-
tions to short-run changes in gasoline prices. However, it is the long-run response
that is the most important in determining which polices are most appropriate for
reducing gasoline consumption. As it turns out, it is relatively difficult to measure
long-run gasoline elasticities in practice due to factors such as the cyclical nature of
gasoline prices. In this paper, we are also limited to currently available data and the
relatively short history of high gasoline prices during the past several years.
Analysis of the short-run price elasticity does however provide some in-
sight into long-run behavior. The long-run response to gasoline price increases is
the sum of short-run changes (miles driven) and long-run changes (fuel economy
of the vehicle fleet). The short-run results suggest that consumers today are less
responsive in adjusting miles driven to increases in gasoline price. This compo-
nent seems unlikely to change significantly for long-run behavior. This is because
factors that may contribute to inelastic short-run price elasticities such as land
use, employment patterns and transit infrastructure typically evolve on timescales
greater than those considered in long-run decisions.

Because that's why.

https://ideas.repec.org/a/aen/journl/2008v29-01-a06.html

Cited by over 300 other papers

What exactly is the point you are trying -- and so far failing -- to make with regard to gasoline price elasticity and road fatalities?

Already explained, and already succeeded. The only two people who failed to understand it are you and gal way cyclist, who are also the two people who have proven themselves to be incapable of understanding the most basic of statistical research. To repeat, that last is a verifiable fact.

Of course, since we know you won't be replying again, because it's -- ahem -- utterly futile, perhaps one of your compadres can answer. No doubt they have it all worked out, and I'm willing to learn from them if not from you.

The only futility is dreaming that you might be capable of learning something new.