Adjusting to 2019 dollars

Hi,

I have pooled multiple single-year ACS files and want to adjust household income, and other income variables, to 2019 dollars. Given that my final dataset consists of pooled single year files, how can I go about adjusting to 2019 dollars? Do I have to divide the cpi multiplier? Any advice is greatly appreciated!

In general to convert from Year X dollars to Year Y dollars, you multiply the Year X dollars by:

CPI (year Y) / CPI (year X)

So for 2017 the annual average CPI-U value (from BLS) is 245.120. For 2019 it is 255.657. To convert 2017 dollars to 2019 dollars, multiply the income amount by 255.657/245.120 = 1.043.

A simple way to do this is to use the variable CPI99 to convert all dollars values to 1999 dollars. Then multiply them all by CPI (2019) / CPI (1999) to convert to 2019 dollars. More info on inflation adjustments in the CPS can be found here - this page is for IPUMS USA, but the variable CPI99 is the same for IPUMS CPS.

I’d like to add something–although the formula above is correct, you don’t have to use the CPI to measure inflation. In fact, you should probably avoid the CPI for data spaced more than 30 or so years apart, since it has many biases that make long-run inflation comparisons inaccurate. The Federal Reserve generally recommends using the PCE price index to measure inflation, as it tends to be less biased over time. You can find it here. Data going back further can be found here.

In re: the following post, which was deleted:

I would like to convert my multiyear wage data from 1970-2018 into 2018 dollar values.
Converting the wage variable to constant 1999 dollar is straightforward (here. In order to convert to 2018 dollars, however, I need the Alternate Factor from table 2. The Factor ends in 2010 in the link above.

I probably did not understand fully @Matthew_Bombyk’s answer above in the second paragraph (mine is same question). CPI2018 = 0.663. If cpi99=1(table 2), then the ratio is just cpi2018.
Can I use a similar table in ipumsCPS for the conversion?

The Alternate Factor in Table 2 at the inflation adjustment page you linked gives the ratio CPIYYYY/CPI1999. I have asked the IPUMS USA team to extend the range of the table at that page. In the meantime, you can take the values from Table 2 at the IPUMS CPS inflation adjustment page, which goes up through 2020. The conversion factors should be exactly the same for CPS and USA (or any other dataset, for that matter).

You can also calculate these yourself by visiting the Bureau of Labor Statistics’ website. Look at the first series “All Urban Consumers (Current Series)”, choose “Top Picks”, and then the first series in the list (“U.S. city average, All items - CUUR0000SA0”). You should then choose to display the value for annual averages. For reference, the CPI2018 value should be 251.107, while the CPI1999 value is 166.6, giving a ratio of 1.507.

The CPI2018 value that you listed (0.663) is the value from the CPI99 variable, and it is actually the inflation multiplier, which is the ratio CPI1999/CPIYYYY. That’s the reciprocal of what you need to use to convert from 1999 dollars to year YYYY dollars.

I agree with @Carlos_Parada about considering alternative inflation adjustment methods, though these are not built into IPUMS USA.

2 Likes

Thank you very much for the detailed response @Matthew_Bombyk, I deleted my post after figuring out by myself and I just see your answer now while browsing. I thought it was not necessary to leave it posted. Apologies if it was wrong.

Thank you also @Carlos_Parada for the comment. The ratio in PCE 2018 dollars/1999 dollars is 1.422 (I am following the same procedure) and that of CPI ratio for the same reference years is 1.507.
Thank you both for the valuable advice.

I am now stuck with why uhrswork is top coded since 1980, and specially for those values of incwage different from N/A or 0. I would really appreciate for any tip on this. I see in this forum that incwage is top coded, but not uhrswork but could not see that in my data. Should I take uhrswork as 99 usual hours worked or do I need to adjust it?

It’s fine to delete your post, I just had already written up a response before I saw it was deleted, so I thought I’d share it anyway. Please see my answer to your other post regarding UHRSWORK.

1 Like