# How does a monthly inflation rate work?

I'm struggling to understand the following inflation numbers for South Africa given here: http://www.tradingeconomics.com/south-africa/inflation-cpi

``````Year  Month  Inflation
2015  Jun    4.7
2015  Jul    5
2015  Aug    4.6
2015  Sep    4.6
2015  Oct    4.7
2015  Nov    4.8
2015  Dec    5.2
2016  Jan    6.2
2016  Feb    7
2016  Mrt    6.3
2016  Apr    6.2
2016  May    6.1
``````

I thought that inflation for this year can roughly be seen as (buying power of my dollar a year ago)/(buying power of my dollar today), and I also thought this would correlate with what services and items cost. For instance, I thought that a chocolate bar's price would roughly follow inflation (i.e. chocolate prices would roughly go up by 5% yearly if inflation is ~5% for those years).

So at first these figures confused me a little, since it surely can't be the "monthly increase"; my buying power couldn't have drop by 4.7% to 7% each month. So I then thought it is a yearly rate given monthly, that this rate should either be divided by 12 to reflect the actual decrease for the month---or be calculates as (wanted rate) = ((given rate)+1)^(1/12)-1 if I must take compounding into account.

But my friend then told me that with the 2015 August and 2015 September rate being both at 4.6 would mean that the prices would most likely not have changed. I don't understand this. What am I missing? Is a larger inflation number for a month not a reflection of how much my money's buying power is falling in that month?

As an expansion on the correct answer:

Consider a really boring economy. Nothing changes; wages and prices stay constant for years at a time. Every month the Consumer Price Index stays at 0%.

Then, something catastrophic happens, say on July 31, 2000. A cheap local source for a vital resource runs out, and it must be obtained from a higher cost source. Floods cut internal road networks, resulting in higher transportation costs. Whatever. The new situation is permanent.

As a result, the next month, August, 2000, prices go up 5%. That is, 5% higher than the previous month, July, 2000, and 5% higher than a year previously, August 1999.

There is a lot of consternation, and politicians each promise that they and only they can wrestle inflation to the ground.

But, when the figures for September, 2000 come out, inflation stays the same. Prices are the same as in August, 2000, and 5% higher than in September, 1999.

This goes on for months. Nothing changes, prices stay the same, and the inflation rate, year over year, stays at 5%.

Finally, the figures for August, 2001 come out. Wonder of wonders; prices are the same as in August, 2000, and inflation drops to zero.

And the politicians all take the credit.

Short version: inflation year over year changes either because of what in now included in the month just past, or what is now excluded from a year ago.

• Thanks, good explanation, when I figured out what the year-on-year monthly inflation actually boils down to, I thought to myself "surely a better way would be to scale previous months with a half-normal distribution". Jul 19, 2016 at 8:43

According to the link you provided, these inflation numbers are the year-over-year inflation amounts for a given month. For example the February number of 7% means from February 2015 to February 2016 the annual inflation rate was 7%.

It's difficult to nail down the impact of inflation at a micro level from month to month. You need to understand what's in the basket of goods. As an example, the price of a widget won't move from \$1 to \$1.04 in one month because of a 4% inflation rate. But considering the incremental price changes of a large number of consumer items over a long time period you can get to an inflation rate. The price of an apple might not have changed but maybe the price of aspirin did.

• Thanks, I just realized this is a "sliding-window" before you answered. Now it makes sense. Jul 15, 2016 at 22:49