6 ms·

Here's the same graph with a logarithmic scale, which is better for comparing rates of change across time: http://oregonstate.edu/cla/polisci/faculty-research/

Here's the same graph with a logarithmic scale, which is better for comparing rates of change across time: http://oregonstate.edu/cla/polisci/faculty-research/sahr/pll1665.htm http://oregonstate.edu/cla/polisci/faculty-research/sahr/pll... And here's the source of both graphs, with links to raw data and more figures: http://oregonstate.edu/cla/polisci/faculty-research/sahr/sahr.htm http://oregonstate.edu/cla/polisci/faculty-research/sahr/sah...

quantumhobbit 17y ago
Thanks for that. Inflation is a cumulative effect and makes a lot more sense on a log scale. Interesting to note that the rate of inflation during the civil war dwarfs anything else, which is what one would expect. I also noticed that there was a huge amount of inflation prior to the Great Depression. The post WWII era is interesting not for the huge amount of inflation but the absence of any deflation.
- thras 17y ago
  No, that's not a graph of inflation rate. That's a graph of price levels. It's not cumulative.
  - mbrubeck 17y ago
    The derivative of the price level (i.e. the slope on these charts) is the inflation rate. The slope on the log-scale chart corresponds to the inflation rate in percentages, which is meaningful. The slope on the linear-scale chart is the inflation rate in "dollars per year" which is not. That's why the log-scale chart is more useful for comparing inflation rates over time.
    - thras 17y ago
      Yes, the derivative is the inflation rate. If we want to talk about that, why not look at a graph of that instead? And assuming we can't for some bizarre reason, the log graph is better at comparing inflation rates, huh? So in which graph do you spot the 1665 to 1945 price stability first?
thras 17y ago
I disagree with the idea that the logarithmic graph is better. It makes it far harder to see the major (and most important) feature of the first graph: a period of relative stability in purchasing power between 1665 and 1945, followed by rapid currency devaluation. Your graph obscures that feature, and replaces it with no useful information. The best piece I've read about the 20th century shift to our current regime of inflation was Dalrymple's recent City Journal article. http://www.city-journal.org/2009/19_3_otbie-inflation.html http://www.city-journal.org/2009/19_3_otbie-inflation.html Sure, inflation is just a kind of wealth redistribution. But wealth redistribution has certain effects. In general, it destroys more than it creates.
- iamwil 17y ago
  "It makes it far harder to see the major (and most important) feature of the first graph" Not if one has learned to read logarithmic graphs. It took me a semester of solid state electronics class (full of log graphs) to get the hang of it, but once you're use to the idea that a straight rising line is exponential, it's easy to see what you say is the most important feature: rapid currency devaluation after 1945. In fact, because it's a log graph, you can tell that it's rising faster than base 10 exponential, because it's still a curve and not a straight line. In addition, because it's a log graph, the large values on the right doesn't dwarf the smaller values on the left. Now, you can see details of what the price change was like locally in time across the board. I'd say there's some useful information there in a log graph. Now if you were saying that log graphs would be dubious to use for the general public, then maybe you'd have something there.
  - thras 17y ago
    I can read logarithmic graphs. And it took longer for my brain to pick out that feature in the second graph. Remember that there is always information trade-off going to a log-scale. In this case, I argue that the trade-off is not worth it. The first graph has a clear and recognizable feature. The second graph is noisy.
- ajross 17y ago
  This is sensationalism. What this graph measures is the move away from asset-backed currency and towards a finance-based central bank. That creates inflation, which is true. It also decouples growth from gold mine output, which is more important. Try graphing GDP per capita across the same range, and note that there's an equivalent jump in the derivative at the same time. Basically, this is a feature, not a bug. Macroeconomics can be terribly dangerous in the wrong hands.
drawkbox 17y ago
This is what I love about HN I was coming here to comment on the fact that all these graphs that attempt to show value are never logarithmic, which is basically completely either dishonest or have huge gaping holes in logic. EVERY graph of anything of value will spike up at the end unless it is logarithmic.

quantumhobbit 17y ago

Thanks for that. Inflation is a cumulative effect and makes a lot more sense on a log scale. Interesting to note that the rate of inflation during the civil war dwarfs anything else, which is what one would expect. I also noticed that there was a huge amount of inflation prior to the Great Depression. The post WWII era is interesting not for the huge amount of inflation but the absence of any deflation.

thras 17y ago
No, that's not a graph of inflation rate. That's a graph of price levels. It's not cumulative.
- mbrubeck 17y ago
  The derivative of the price level (i.e. the slope on these charts) is the inflation rate. The slope on the log-scale chart corresponds to the inflation rate in percentages, which is meaningful. The slope on the linear-scale chart is the inflation rate in "dollars per year" which is not. That's why the log-scale chart is more useful for comparing inflation rates over time.
  - thras 17y ago
    Yes, the derivative is the inflation rate. If we want to talk about that, why not look at a graph of that instead? And assuming we can't for some bizarre reason, the log graph is better at comparing inflation rates, huh? So in which graph do you spot the 1665 to 1945 price stability first?

thras 17y ago

I disagree with the idea that the logarithmic graph is better. It makes it far harder to see the major (and most important) feature of the first graph: a period of relative stability in purchasing power between 1665 and 1945, followed by rapid currency devaluation. Your graph obscures that feature, and replaces it with no useful information. The best piece I've read about the 20th century shift to our current regime of inflation was Dalrymple's recent City Journal article. http://www.city-journal.org/2009/19_3_otbie-inflation.html http://www.city-journal.org/2009/19_3_otbie-inflation.html Sure, inflation is just a kind of wealth redistribution. But wealth redistribution has certain effects. In general, it destroys more than it creates.

iamwil 17y ago
"It makes it far harder to see the major (and most important) feature of the first graph" Not if one has learned to read logarithmic graphs. It took me a semester of solid state electronics class (full of log graphs) to get the hang of it, but once you're use to the idea that a straight rising line is exponential, it's easy to see what you say is the most important feature: rapid currency devaluation after 1945. In fact, because it's a log graph, you can tell that it's rising faster than base 10 exponential, because it's still a curve and not a straight line. In addition, because it's a log graph, the large values on the right doesn't dwarf the smaller values on the left. Now, you can see details of what the price change was like locally in time across the board. I'd say there's some useful information there in a log graph. Now if you were saying that log graphs would be dubious to use for the general public, then maybe you'd have something there.
- thras 17y ago
  I can read logarithmic graphs. And it took longer for my brain to pick out that feature in the second graph. Remember that there is always information trade-off going to a log-scale. In this case, I argue that the trade-off is not worth it. The first graph has a clear and recognizable feature. The second graph is noisy.
ajross 17y ago
This is sensationalism. What this graph measures is the move away from asset-backed currency and towards a finance-based central bank. That creates inflation, which is true. It also decouples growth from gold mine output, which is more important. Try graphing GDP per capita across the same range, and note that there's an equivalent jump in the derivative at the same time. Basically, this is a feature, not a bug. Macroeconomics can be terribly dangerous in the wrong hands.

drawkbox 17y ago

This is what I love about HN I was coming here to comment on the fact that all these graphs that attempt to show value are never logarithmic, which is basically completely either dishonest or have huge gaping holes in logic. EVERY graph of anything of value will spike up at the end unless it is logarithmic.