In a recent post, I surveyed indirect cost rates across 50 institutions. In the course of the data analysis related to this post, a colleague pointed out that indirect cost information is available in NIH RePORTER. This surprised me since it did not used to be available and, indeed, further checking revealed that such data are only available for FY2012 and beyond. Nonetheless, the availability of these data does enable analysis of indirect cost expenditures aggregated over the NIH portfolio (as opposed to indirect cost rate data).

I downloaded such direct and indirect cost data for all R01 grants for FY2013 and calculated the ratio of indirect costs (IDC) to direct costs (DC) for all grants with direct costs over $10,000 and with non-zero indirect costs. These 23773 grants are broken down in three groups: Type 5 (non-competing renewal) grants, Type 1 (new) grants, and Type 2 (competing renewal) grants. For this analysis, I did not include supplements (Type 3) or other types. The IDC/DC ratios are plotted below. Also shown are lines corresponding to ratios of 0.40, 0.50, and 0.60.

The ratios of total indirect costs to total direct costs are 0.473, 0.469, and 0.514 for Type 5, Type 1, and Type 2 grants, respectively. For the whole collection, this ratio is 0.475. Note that indirect costs on a given grant may be less than that expected based on the indirect cost rates because of exclusions such as equipment or, less frequently, greater due to subcontracting costs.

Given the small differences in overall IDC/DC ratios for the three Types of R01 grants, how do the distributions of IDC/DC ratios differ? These distributions are shown below.

These distributions are quite similar. There does appear to be a smaller fraction of Type 2 grants with relatively low IDC/DC ratios compared with Type 5 and Type 1 grants. This may reflect differences in the institutions from which investigators typically succeed in renewing RO1 grants, although this hypothesis will require further analysis.

Some commenters have suggested that a disproportionate fraction of grants going to a set of institutions with relatively high indirect cost rates could distort the overall distribution of funds. As a model of this behavior, suppose that grants were ranked from the highest IDC/DC ratio to the lowest IDC/DC ratio and that the top (highest IDC/DC) grant had direct costs 0f 1.2 times the average and the bottom (lowest IDC/DC) grant had direct costs 0f 0.8 times with average with grants in between funded at intermediate values linearly varying between 1.2 and 0.8 of the average. To analyze this distribution, let us plot the cumulative total funding versus the rank from highest to lowest IDC/DC as shown below.

The plot shows that the curve is concave down. The dotted red line shows the expectation with no differences between grant amount at different IDC/DC ratios (i.e. a straight line).

How does the FY2013 distribution compare with this simulation? The corresponding plot is shown below:

In contrast to the simulation, the curve is nearly straight for high values of the IDC/DC ratio and then slightly concave up (rather than down) for lower values. Although there are certainly other ways to analyze this distribution, this first pass analysis does not support the notion that the distribution of a disproportionate amount of funding going to a subset of institutions with relatively high IDC rates results in a substantial increase in overall indirect cost expenditures compared with other distributions.

As I noted above, data are also available for FY2012. It will be interesting to see if there are any significant differences between FY2013 and FY2012 although I will be surprised if there are. In addition, I am trying to find out if NIH can release data for a wide range of years so that trends over time could be examined.

The notion that a few rich institutions can skew the curve is a non-starter when you see how little variability there is in the data set overall. Even including the 3 or 4 "outliers", the overall mean IDC is 58.184 +/- 7.771 (StDev), SEM = 1.122. For this argument to hold, you'd need a couple dozen places with IDCs in the 70s and above.

Still though, "eat the rich" does have a nice ring to it.

I think it's pretty clear at this point that the problem is too many mouths at the trough. The real question to ask about IDC's is not "If we cut IDC rates can we all have more grants?", but "Are IDC's structured in such a way that they encourage unsustainable growth?"

Or perhaps "Should IDCs be restructured to encourage more sustainabile growth and, if so, in what ways, and how should key stakeholders be engaged in such a discussion?"

Ola- well observed. This joins the dataset on the 5-R01 PIs to show how pointing the finger at a tiny fraction of the distribution can't possibly fix the systemic problems. Also joins the analysis showing how evening up the success rates of low-representation PIs would barely budge the needle on majority-identity PI success rates. The NIH enterprise is very large and we are all subject to cherry picking. Congrats once again to datahound for making us face the full distribution.

datahound- would there ever be anyway to figure out the degree to which Universities take advantage of charging their capital infrastructure costs to IDC? Or is this now universal practice?

Is it “eat the rich”, or “the rich eat”?

[…] The Effective Rate drops from 44.2% to a low of 37.2% in 2012 before rising slightly over the past two years. These values are all somewhat lower than I anticipated based on my previous analysis on R01s. […]