In the tax credit program, rents are tied to income limits published by the U.S. Department of Housing and Urban Development (HUD). The current list of these limits tells you the maximum rent that can be charged in each unit in the project at present. A conventional underwriting assumption is that rental income will increase at 2% a year.

However, as someone with tax credit experience, you know that income limits and tax credit rents have remained frozen in some parts of the country in the recent past, so you decide to look at the underlying numbers more closely before adopting the conventional 2% assumption.

You see that the applicable income limit in the Johnson City metro area is the same in 2008 as it was in 2007, so the rent ceilings in tax credit units are also the same.  You also see that HUD’s AMI estimates for the Johnson City metro over the past three years are as follows:

     2006   $44,900
     2007   $49,800
     2008   $44,700

In other words, the AMI estimate for your area actually dropped considerably in 2008.  When this happens, HUD typically applies a “hold harmless” policy that freezes income limits (and tax credit rent ceilings) rather than let them decline from one year to the next.  Rents won’t go down, but it’s unlikely they’ll rise any time soon.

2008 Brings Frozen Income Limits

How many areas of the country find themselves in this situation in 2008?  The areas for which HUD produces AMI estimates and income limits include all metropolitan areas and all non-metropolitan counties in the country. Because there are more non-metropolitan counties than metro areas, but the bulk of the U.S. population resides in metro areas, it makes sense to track the two groups separately. On the current list released on 2/13/2008, income limits were frozen in 190 of 532 metro areas (roughly 36%), and in 637 of 2,043 non-metro counties (31%).  Although down from 2007, this is still a large number compared to the number of areas where income limits are frozen in most years (Figure 1).

The figure also indicates the years when shocks (the introduction of new data sets, changes in methodology etc.) hit the system used to estimate AMI and setting income limits. There were a number of shocks since 2002. The largest occurred in 2007, when HUD adopted a methodology that essentially benchmarks a large fraction of the areas in the country to new local data that became available when the Census Bureau fully implemented its American Community Survey (ACS). As the figure shows, this caused AMI estimates to decline, and income limits to be frozen across about two-thirds of the country. 

The 2007 change was so extreme that, to some, the situation in 2008 may appear relatively mild. However, the number of areas with frozen income limits in 2008 appears very large compared to the more normal years at the beginning of the decade, when the estimates were not being affected by changes in data, definitions, or methodology (Figure 2).

There Are Areas Where Income Estimates Declined

The AMI reductions for many areas in 2007 were large enough to cause income limits that were frozen in that year to remain frozen in 2008, even though AMI estimates may have increased.  (In order to “unfreeze” income limits, the AMI must not only rise, but rise far  enough to break above its peak in any previous year). Thus, of the 190 metro areas with frozen income limits in 2008, it’s not surprising that 170 also had frozen limits in 2007. Similarly, of the 637 non-metro counties with frozen income limits in 2008, 629 also had frozen limits in 2007. In these places, income limits and tax credit rent ceilings in 2008 are the same as they were in 2006. In some cases, the freeze has been much longer. The counties that remained in the Atlanta metro area throughout the decade, for instance, provide an example of a large metropolitan area where income limits and rents have not increased at all since 2002.

Although many large reductions in AMI occurred in 2007, there are nevertheless more than a hundred cases where ACS-based estimates of AMI declined in 2008. Table 1 lists the areas with the largest declines on a percentage basis. Because of the hold-harmless policy, income limits and tax credit rent ceilings were frozen in these areas in 2008.  Only three of the entries in Table 1—New Bedford, Mass., Cleveland, Tenn., and Reading, Pa.—also had income limits that were frozen in 2007. The sharp reductions in AMI, coming on the heels of income limit freezes for the previous period, raise questions about when stakeholders in these areas may reasonably expect income limits to rise again in the future.

And, There Are Areas Where Income Estimates Increased

There are reasons to look at the other end of the spectrum—that is, at areas of the country the that experienced large increases in AMI 2008. For those trying to evaluate the methodology adopted in 2008, two consecutive years of fully implemented ACS data provide the first opportunity to see how the new approach causes AMI estimates to fluctuate up as well as down. 

The most extreme case of an upward fluctuation in 2008 is the Texarkana metro area in Texas and Arkansas, where HUD’s estimate of AMI increased by more than 20%. The top ten areas according to this measure all experienced AMI increases of more than 13% (Table 2).

Although rising incomes may signal an area of new development potential, this is only true if the rising incomes are not a statistical artifact. The one-year increases shown in the table are large enough to suggest that they may be due to statistical error rather than a real, underlying trend.  

HUD takes steps to account for statistical error and smooth the data.  The “hold harmless” policy that holds income limits constant, rather than allowing them to drop in one year and rebound the next, is one way to smooth the series. HUD also incorporates state level ACS data into its local AMI estimates, weighting the state data more heavily when the local ACS data have a larger margin of error. In the case of the Texarkana metro area, the local ACS data indicated that AMI jumped by more than 30%, but the data-smoothing procedures reduced this to a 20% increase.

Income Limits Did Rise in Some Areas

By themselves, large increases in AMI don’t always translate into equivalent increases in income limits and tax credit rent ceilings. Although the Very Low Income Limit (VLIL) that applies to the tax credit program is roughly based on 50% of AMI, a large number of VLILs are not  actually equal to this percentage in practice.  In addition to the “hold harmless” policy, HUD applies a number of other adjustments that alter the AMI/VLIL relationship.

Table 3 shows the areas that experienced the largest increases in VLIL in 2008, and also provides information on the basis for the VLILs. In only one case—the Billings, Montana metro area—was the VLIL based strictly on 50% of AMI in both 2007 and 2008. Texarkana is the only area that appears on both Tables 1 and 2, and it is on the top of Table 2 but the bottom of Table 3. (Billings narrowly missed making both tables: it would be the thirteenth area listed in Table 2, if the table extended that far.)

Many of the increases in Table 3 are not due at all to changes in the underlying AMI, but are based on a “housing cost” adjustment HUD employs.  If the VLIL is below a certain multiple of the area’s Fair Market Rent (FMR), HUD raises the VLIL up to that multiple (a similar adjustment is used to reduce the VLIL in a few cases). The method for computing FMRs changed in 2008, and many local FMRs also are now based largely on ACS data. This change caused FMRs to increase substantially in several areas. Table 3 includes cases like Honolulu, where the housing cost adjustment applied in both 2007 and 2008; Santa Barbara, where the VLIL was set at half of AMI in 2007 but raised to a multiple of the FMR in 2008; and Monroe County FL, where the VLIL was frozen in 2007 but became unfrozen in 2008 as the result of a large increase in the area’s FMR.

Thus, the large number of areas with frozen FMRs in 2008 occurred despite the increases in FMRs (probably a one-time event, attributable to the change in FMR methodology) that had a tendency to unfreeze them.

Meanwhile, Back in Johnson City…

Returning to the example of a stakeholder in an existing or planned tax credit project in the Johnson City metro area—what would be a reasonable and prudent estimate of the project’s rental income over the next several years? No matter what happens to inflation or operating costs, the maximum rents that can be charged will not increase until the AMI estimate breaks above the $49,800 established in 2007.

If the current AMI estimate increased at a uniform rate of 2% a year, this wouldn’t happen until 2014, so maybe you should assume no nominal growth (and therefore real declines) in rental income for the project until then.

However, the estimate of AMI in this metro area has not been exhibiting anything close to a constant and predictable growth rate. Maybe the relatively high AMI in 2007 was a statistical anomaly, and the AMI that was nearly the same in 2006 and 2008 better represents the actual trend, in which case nominal rent growth would not be expected to until far past 2014.

On the other hand, you could argue that the AMI in 2006 should be the anomaly, due to the substantial change in methodology that took place in 2007 and should make AMIs for 2007 and 2008 more comparable. Yet, despite a consistent methodology based on a full year of ACS data in both cases, the 2007 and 2008 numbers differ by more than 10%. 

How Do We Fix This?

The bottom line is that these fluctuations in AMI, which so far seem to persist even when consistent methodologies and data sets are employed, make it difficult  to underwrite tax credit projects with any of the techniques or assumptions that have become standard in the industry. As a result, NAHB is advocating changes in the way income limits and rent ceilings are set in the Low-Income Housing Tax Credit (LIHTC) program, as well as in closely related programs like tax-exempt bonds and HOME. The current draft of a tax credit modernization bill in the House seeks to address this to some extent.

For more information on that effort, e-mail Carmel McGuire at NAHB Multifamily’s Housing Credit Group, or call her at 800-368-5242 x8207.
[ return to top ]