US price levels 1665 to 2005, with 2005 = 100 | Hacker News Reader

US price levels 1665 to 2005, with 2005 = 100

oregonstate.edu

13 pointsAnon8424 comments

Open articleView on HN

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/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/sah...

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.

No, that's not a graph of inflation rate. That's a graph of price levels. It's not cumulative.

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.

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?

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

Sure, inflation is just a kind of wealth redistribution. But wealth redistribution has certain effects. In general, it destroys more than it creates.

"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.

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.

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.

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.

I don't quite get it. What is this data based on? What goods are included?

The price of food, for instance, has fallen dramatically just in the past 50-75 years. In both percent-of-household-budget and real dollar terms, it's dirt cheap compared to what it used to be. So any CPI-type "basket" that includes food ought to show price deflation over time.

In the meantime, many things that would have been unthinkable luxuries have become regarded as necessities of one sort or another. I suspect the average number of square feet of living space has gone up dramatically, as has energy consumption. But to just factor that into "price" doesn't seem particularly fair, since the 'average' standard of living in 2005 is so much higher than in 1795.

The chart is neat but it needs more context.

I completely agree, except for the last line. This chart is worthless without any details as to how the numbers are calculated.

> and real dollar terms

What does that mean? Obviously, you need more US dollars today to purchase food than you did in 1950, due to inflation. Does "real dollar terms" mean adjusted for inflation?

http://www.gti.net/mocolib1//prices/1950.html

1950: Beef is $.43/lb

http://www.gti.net/mocolib1//prices/2009.html

2009: Beef is $2.99/lb

"real dollars" means something like "percentage of disposable income".

Also it's impossible to make a meaningful comparison of things you buy for money today compared to things that were bought for money two hundred years ago. Back then a lot more people grew their own food, for example. And they didn't work for software startups and buy all their tech equipment from minimum wage factories in China.

Yes, basically. "Real dollars" are dollars adjusted for inflation.

I think it's a little more precise than just saying "inflation-adjusted dollars" because when people hear that, they automatically assume "adjusted using CPI," which is only one way to try and correct for inflation. It's a reasonably well-accepted way, but there are others. (E.g.: labor-hours are popular in some circles.)

Cf. http://en.wikipedia.org/wiki/Real_versus_nominal_value_%28ec...

This graph seems off to me. I have a hard time believing the 80 years from 1665 to 1745 were a period of constant deflation.

If you look carefully, it's not constant deflation in that period -- over those 80 years there's an average deflation rate of 0.6%, but at many points the price index jumps 10% above the trendline.

I find these numbers to be entirely believable -- after all, this is a period when the American Colonies are expanding rapidly, with resulting effects in terms of specialization of skills and self-sufficiency. It's going to be far cheaper to have something produced in 1745 by the local expert X-maker than it was to import the same item from Britain in 1665.

Boy is that not the story here. It's what happens after WWII. There is pretty much no measure of currency that will change this basic story, which has two parts: USD as a reserve currency post Bretton Woods, and abandoning the gold standard giving notes no intrinsic value.

Gold has no intrinsic value either; its value is barter value, what other people believe it's worth.

Fiat money is no different; its value is barter value, what other people think it's worth, which includes how much they trust the issuer not to debase it.

Hypothesis: possible partial cause of the sudden spike ~ 1945 - 1960 where the graph takes off:

Introduction of television into mainstream society. Fueled by advertising, everything "costs" more: entertainment is fueled by advertising. News is also fueled by advertising. Advertising places a new layer of "value" on products, causing the sudden perceived need to pay more.

Interesting data.

The graph "takes off" where it does because economists figured out that deflation harms the economy by encouraging irrational hoarding of money. Modern central banks aim to keep inflation predictable, low, and positive; several banks have explicit targets (usually 2%) which are chosen largely by asking the question "given that inflation rates have inherent volatility, what's the lowest average inflation rate we can have without volatility regularly pushing it below zero?"

This is true; regarding this specific time period, however, there were definitely some large historical events that contributed to prices rising. The logarithmic scale reveals it as well. Vietnam, yes, but also that whole patriotism thing, news reporting, and "Buy American" which was, obviously, fueled by the birth of the advertising age.

I made an interactive version of the chart with his later data (that now starts in 1774) here http://verifiable.com/charts/3937 (I also added the log version so you can compare them)

Is this supposed to be proving any sort of a point?

Serious question; what's the point here?

Reply on news.ycombinator.com