All Things Considered: February 2011
The man at the front of the lecture theatre was cool, relaxed, self-assured and very funny. He was also taking a jackhammer to the foundation of many economists’ stock in trade – economic models.
The man at the front of the lecture theatre was cool, relaxed, self-assured and very funny. He was also taking a jackhammer to the foundation of many economists’ stock in trade – economic models.
Speaking to two dozen experienced business and finance reporters he, as I recall, said: “The one thing you should know is that all models are flawed.” Some are worse than others, but none produces results you would want to take to the bank, he said.
The man in question wasn’t an economist; he was a statistician who made a living dissecting economic and other models for governments, industry and the military. He looked for margins of error in data and calculations; he tested assumptions and their implications. In an intense two-week program designed to bring reporters up to speed on economic thought he was the heretic brought in to caution against treating theory as truth and projections generated by economic models as fact.
Unfortunately, that is what happens. A study commissioned by the European Commission and the Canadian government predicts $12 billion in annual economic gains for Canada from a free trade agreement with Europe and that figure is repeated as fact in virtually every news report on the negotiations.
For supply-managed industries this may be a problem.
It would be helpful if that $12 billion were not treated as fact but as the statistician said, merely an inevitably flawed projection.
He pointed out that models are often based on imperfect data with margins of error. Even if the errors are tiny, if you pile tiny error on top of tiny error it may all come out in the wash or they may cascade into chaos. Lately, the world has seen what that chaos can look like. None of the mainstream economic models forecast what is now called The Great Recession.
Even worse are erroneous assumptions. Joseph Stiglitz, the 2001 Nobel Prize winner in economics, took on one assumption common to many, if not most, economic models – the assumption of perfect information. Perfect information is assumed in models because (a) it fits the theory and (b) you have to assume it or the models don’t work.
“I only varied one assumption – the assumption concerning perfect information – and in ways which seemed highly plausible,” he wrote in his Nobel presentation. “Changing only one assumption in ways which were totally plausible had drastic consequences,” he wrote.
Jim Stanford, an economist with the Canadian Auto Workers, has taken a look at the economic arguments behind the proposed Canada-EU trade agreement and found them wanting.
In his study “Out of Equilibrium: The Impact of EU-Canada Free Trade on the Real Economy,” which has been published by the Canadian Centre for Policy Alternatives, he writes: “this finding (of $12 billion in economic gain) relies upon extreme and far-fetched assumptions regarding the self-adjusting nature of all markets, and the manner in which free trade would be implemented and experienced.”
In effect, he says, the assumptions in the model determine the outcome it will produce. “The findings of the EU-Canada study amount to an assertion that free trade will produce mutual economic gains, not a demonstration that this will be the case,” he writes.
Stanford then presents three alternative scenarios: one in which tariffs are mutually eliminated (which ought to scare the heck out of supply-managed industries); one in which EU-Canada trade expands in line with the historical experience of Canada’s previous FTAs; and one in which tariff elimination is combined with the appreciation of Canada’s currency (versus the euro), which has already happened.
“In every case, the bilateral trade balance worsens significantly (and in the third scenario, it worsens dramatically – since the higher Canadian dollar reduces Canadian exports, even as imports from the EU are surging).”
“The simulations suggest an incremental loss of between 28,000 jobs (in the first scenario) and 150,000 jobs (in the third). Direct losses in Canadian GDP range between 0.56 percent in the first scenario, and almost three percent in the third,” he writes.
The scenarios presented by Stanford are unlikely to evolve as forecast. The assumptions may be flawed; the data may contain errors. But they may contain fewer flaws and errors than the government study. At a minimum they are a counterpoint to the assertion that the proposed trade deal is win-win. If nothing else, Stanford’s work should raise questions.
Jim Stanford’s complete report can be found at: www.policyalternatives.ca.
Speaking to two dozen experienced business and finance reporters he, as I recall, said: “The one thing you should know is that all models are flawed.” Some are worse than others, but none produces results you would want to take to the bank, he said.
The man in question wasn’t an economist; he was a statistician who made a living dissecting economic and other models for governments, industry and the military. He looked for margins of error in data and calculations; he tested assumptions and their implications. In an intense two-week program designed to bring reporters up to speed on economic thought he was the heretic brought in to caution against treating theory as truth and projections generated by economic models as fact.
Unfortunately, that is what happens. A study commissioned by the European Commission and the Canadian government predicts $12 billion in annual economic gains for Canada from a free trade agreement with Europe and that figure is repeated as fact in virtually every news report on the negotiations.
For supply-managed industries this may be a problem.
It would be helpful if that $12 billion were not treated as fact but as the statistician said, merely an inevitably flawed projection.
He pointed out that models are often based on imperfect data with margins of error. Even if the errors are tiny, if you pile tiny error on top of tiny error it may all come out in the wash or they may cascade into chaos. Lately, the world has seen what that chaos can look like. None of the mainstream economic models forecast what is now called The Great Recession.
Even worse are erroneous assumptions. Joseph Stiglitz, the 2001 Nobel Prize winner in economics, took on one assumption common to many, if not most, economic models – the assumption of perfect information. Perfect information is assumed in models because (a) it fits the theory and (b) you have to assume it or the models don’t work.
“I only varied one assumption – the assumption concerning perfect information – and in ways which seemed highly plausible,” he wrote in his Nobel presentation. “Changing only one assumption in ways which were totally plausible had drastic consequences,” he wrote.
Jim Stanford, an economist with the Canadian Auto Workers, has taken a look at the economic arguments behind the proposed Canada-EU trade agreement and found them wanting.
In his study “Out of Equilibrium: The Impact of EU-Canada Free Trade on the Real Economy,” which has been published by the Canadian Centre for Policy Alternatives, he writes: “this finding (of $12 billion in economic gain) relies upon extreme and far-fetched assumptions regarding the self-adjusting nature of all markets, and the manner in which free trade would be implemented and experienced.”
In effect, he says, the assumptions in the model determine the outcome it will produce. “The findings of the EU-Canada study amount to an assertion that free trade will produce mutual economic gains, not a demonstration that this will be the case,” he writes.
Stanford then presents three alternative scenarios: one in which tariffs are mutually eliminated (which ought to scare the heck out of supply-managed industries); one in which EU-Canada trade expands in line with the historical experience of Canada’s previous FTAs; and one in which tariff elimination is combined with the appreciation of Canada’s currency (versus the euro), which has already happened.
“In every case, the bilateral trade balance worsens significantly (and in the third scenario, it worsens dramatically – since the higher Canadian dollar reduces Canadian exports, even as imports from the EU are surging).”
“The simulations suggest an incremental loss of between 28,000 jobs (in the first scenario) and 150,000 jobs (in the third). Direct losses in Canadian GDP range between 0.56 percent in the first scenario, and almost three percent in the third,” he writes.
The scenarios presented by Stanford are unlikely to evolve as forecast. The assumptions may be flawed; the data may contain errors. But they may contain fewer flaws and errors than the government study. At a minimum they are a counterpoint to the assertion that the proposed trade deal is win-win. If nothing else, Stanford’s work should raise questions.
Jim Stanford’s complete report can be found at: www.policyalternatives.ca.