4 ms·

I'm of the opinion that economic statistics become almost worthless from the perspective of politics and policy once you get to a certain sample size. A cost o

by api 8y ago

I'm of the opinion that economic statistics become almost worthless from the perspective of politics and policy once you get to a certain sample size. A cost of living statistic that averages out cost of living between say Michigan and coastal California is quite wrong in both places. It will grossly underestimate cost of living in California and overestimate it in Michigan. I think this is one of the major problems with politics and economic policy setting at the national and global level. National and global leaders look at big aggregate statistics, which completely blinds them to the street level realities everywhere except those places that happen to coincide with the statistical median. Most places do not coincide with the statistical median. I think this largely explains the recent global populist uprising as well as the fact that global leaders were surprised by it. Large scale macro statistics might be useful for setting things like Fed interest rates, but they're nearly worthless for addressing local problems.

anoncoward111 8y ago
This is exactly correct. Additionally, qualitative data needs to be factored into policy decisions as well. If a city looks and feels unsafe, dirty, deserted, and so on, then it probably is those things. No amount of statistics can explain away the experience of any unbiased, reasonable observer. Park your car in the center of town somewhere. Hang out and eat a sandwich with your windows open in your car. See who walks by, hear the noises, the beeps, the sirens, the shouting kids or quiet wind. Then you'll know what a town is really like.
- IkmoIkmo 8y ago
  Not sure I agree there. There are plenty of historical examples of people seeing things in bizarre ways that simply weren't true. We see this recently with crime statistics dropping across the board, terrorism being less dangerous than bees, and people feeling less safe than ever. I'd say plenty of people aren't fully aware of ways in which their lives get cheaper or better. That's just 'normal'. Whereas growing up and paying $2 for a coffee, and now 15 years later it's $3, is just a sign of you getting squeezed, and people are very much aware of that. One cognitive / measurement issue here is hedonic inflation. For example, the average new home was 1000 square feet in the 20th century and 2500 square feet today. Meanwhile, average household sizes decreased. We all got bigger houses with fewer people to share them with. We also got universal central heating, hot piped water, flush toilets. Similarly, our computers have gotten faster. I don't need to explain the implications of moore's law to anyone here. Suppose the average $1000 computer gets twice as fast previous year. Is that to be measured as inflation or no change? No change? Perhaps, average prices remained constant. But inflation is, among other things, a purchasing power kind of metric. 'How much computing power can I purchase with this money'. In that sense, prices halved. In short, there was deflation, on a hedonic basis. This is something a lot of people don't realise. Inflation figures are sometimes suppressed by such improvements (which halved computing prices, even if sticker prices didn't change). Homes got more than twice as big, computers a thousand times as fast, these things can't simply be ignored by stating 'average homes and computers are more expensive', which is the reality that people experience in real life when looking at sticker prices of computers and rental properties. The reality that people often omit is that the product itself has massively improved. Same with healthcare. Yes, costs increased. So did outcomes. Try surviving cancer today vs 50 years ago. Try living till 80 years old on average, 50 years ago, when life expectancy was sub 70. https://ourworldindata.org/grapher/life-expectancy?tab=chart&country=USA https://ourworldindata.org/grapher/life-expectancy?tab=chart...
syntaxfree 8y ago
There's also Goodhart's law: when a measure becomes a target, it ceases to be a good measure.
- api 8y ago
  You see this a lot in machine learning. Reward a neural network or evolutionary algorithm for identifying squirrels vs. cats and you'll get one that compares one pixel of the usually beige wall color in mostly indoor shots of cats vs. the non-beige pixel found in mostly outdoor shots of squirrels.
IkmoIkmo 8y ago
Fortunately we have exactly such local and sector level inflation statistics. They just don't get published in the media much because who wants to read tables for all states and provinces of tens of different countries. (e.g. https://www.bls.gov/regions/subjects/consumer-price-indexes.htm https://www.bls.gov/regions/subjects/consumer-price-indexes....) The national government looks at national figures, that's not worthless. The state looks at state figures. And the municipality looks at local figures. This idea that politicians are completely blind and have no idea of street level realities, only exist at the national level, or have no data or policy instruments beyond the national level is a joke, but it seems almost like you're implying just that.

by api 8y ago

anoncoward111 8y ago

This is exactly correct. Additionally, qualitative data needs to be factored into policy decisions as well. If a city looks and feels unsafe, dirty, deserted, and so on, then it probably is those things. No amount of statistics can explain away the experience of any unbiased, reasonable observer. Park your car in the center of town somewhere. Hang out and eat a sandwich with your windows open in your car. See who walks by, hear the noises, the beeps, the sirens, the shouting kids or quiet wind. Then you'll know what a town is really like.

IkmoIkmo 8y ago
Not sure I agree there. There are plenty of historical examples of people seeing things in bizarre ways that simply weren't true. We see this recently with crime statistics dropping across the board, terrorism being less dangerous than bees, and people feeling less safe than ever. I'd say plenty of people aren't fully aware of ways in which their lives get cheaper or better. That's just 'normal'. Whereas growing up and paying $2 for a coffee, and now 15 years later it's $3, is just a sign of you getting squeezed, and people are very much aware of that. One cognitive / measurement issue here is hedonic inflation. For example, the average new home was 1000 square feet in the 20th century and 2500 square feet today. Meanwhile, average household sizes decreased. We all got bigger houses with fewer people to share them with. We also got universal central heating, hot piped water, flush toilets. Similarly, our computers have gotten faster. I don't need to explain the implications of moore's law to anyone here. Suppose the average $1000 computer gets twice as fast previous year. Is that to be measured as inflation or no change? No change? Perhaps, average prices remained constant. But inflation is, among other things, a purchasing power kind of metric. 'How much computing power can I purchase with this money'. In that sense, prices halved. In short, there was deflation, on a hedonic basis. This is something a lot of people don't realise. Inflation figures are sometimes suppressed by such improvements (which halved computing prices, even if sticker prices didn't change). Homes got more than twice as big, computers a thousand times as fast, these things can't simply be ignored by stating 'average homes and computers are more expensive', which is the reality that people experience in real life when looking at sticker prices of computers and rental properties. The reality that people often omit is that the product itself has massively improved. Same with healthcare. Yes, costs increased. So did outcomes. Try surviving cancer today vs 50 years ago. Try living till 80 years old on average, 50 years ago, when life expectancy was sub 70. https://ourworldindata.org/grapher/life-expectancy?tab=chart&country=USA https://ourworldindata.org/grapher/life-expectancy?tab=chart...

syntaxfree 8y ago

There's also Goodhart's law: when a measure becomes a target, it ceases to be a good measure.

api 8y ago
You see this a lot in machine learning. Reward a neural network or evolutionary algorithm for identifying squirrels vs. cats and you'll get one that compares one pixel of the usually beige wall color in mostly indoor shots of cats vs. the non-beige pixel found in mostly outdoor shots of squirrels.

IkmoIkmo 8y ago

Fortunately we have exactly such local and sector level inflation statistics. They just don't get published in the media much because who wants to read tables for all states and provinces of tens of different countries. (e.g. https://www.bls.gov/regions/subjects/consumer-price-indexes.htm https://www.bls.gov/regions/subjects/consumer-price-indexes....) The national government looks at national figures, that's not worthless. The state looks at state figures. And the municipality looks at local figures. This idea that politicians are completely blind and have no idea of street level realities, only exist at the national level, or have no data or policy instruments beyond the national level is a joke, but it seems almost like you're implying just that.