Consider a pair of very dramatic contrasts. At the end of April, the Virginia DOT announced a development agreement with the Fluor/Transurban coalition under which their planned addition of four HOT lanes to the Washington Beltway will go forward, supported 100% by the toll revenues generated by the new lanes. By contrast, the feasibility study of a network of HOT lanes for Atlanta (a metro area whose annual congestion cost, as measured by Texas Transportation Institute, is nearly as great as that of the DC metro area), concluded that, at best, HOT lane revenues can cover operating and maintenance costs but only the incremental capital costs of doing the lanes as HOT rather than HOV (i.e., none of the actual lane construction costs).
The other contrast is between Denver and Minneapolis/St. Paul. The feasibility study of a network of express toll lanes in Denver concluded that toll revenues could cover at least 50-60% of the capital costs of a $4.8 billion system. But the study of a $3.5 billion network of (mostly) express toll lanes in the Twin Cities found that toll revenues could cover an average of 22% of capital costs. And note well that the annual congestion cost, as calculated by TTI, is almost identical for these two metro areas. That should be an indication of the strength of underlying demand for congestion relief. But the study results are surprisingly different.
I've spent some time going through the Atlanta, Denver, and Twin Cities studies, attempting to figure out why the findings are so different. It's not easy going, since they were done by different firms, all highly qualified in transportation modeling, and the studies are quite detailed. Here is what I've learned so far.
My first thought was that the difference might be accounted for by different policies toward free passage by HOVs. If you give away the valuable space in costly-to-build HOT lanes to two-person car-pools (HOV-2), there will be little space left to sell to paying customers. But that hypothesis did not hold up. The Atlanta study results I cited above were based on the HOV-4+ version, under which only HOVs with four or more occupants would get free passage. The Twin Cities study included a couple of converted HOV lanes which would let HOV-2s go free, but all the new construction (the bulk of the network) would be express toll lanes with no HOV freebies. Denver was modeled as all-ETL. And by contrast, the 100% self-supporting Beltway HOT lanes would allow HOV-3s (of which there are many in the area) to go free.
My second hypothesis was that the studies made different assumptions about HOT lane toll rates. This idea did explain some of the differences. The Georgia study used toll rates of between $.02 and $.14 per mile, clearly far below what experience has shown to be the market-clearing price for congestion relief, at least in Southern California (on I-15 and SR 91). The Minnesota study used three rates—$.10, $.30, and $.50—but seemed to use the two higher ones very sparingly. The Denver study used a complete range of $.05 to $.50, and seemed more realistic in applying the higher rates.
And here's another crucial difference. The Atlanta study appears not to have adjusted its toll rates for inflation. This makes no sense, since value pricing depends on keeping the price at a market-clearing level on into the future (which means at least annual CPI increases). Traditional toll roads were often financed based on constant toll rates to cover constant annual debt service, relying on growth in traffic volume to increase the revenue over time. But a HOT lane's revenue profile over time is dramatically different. This can be seen in tables in the Denver study, showing, for each freeway segment with an ETL, the projected annual revenue. For example, I-25 ETLs (Scenario 1) are forecast to generate $15 million in 2010 and $93 million in 2040; thatï¿½s a lot more than could be generated simply by growth in traffic volume!
The studies also used somewhat different definitions of financial feasibility. The Atlanta study compared the annualized capital costs with annual toll revenue in 2030. The Twin Cities study compared the present value of the revenue stream (2008 to 2030) with the present value of capital costs. But the Denver study did something similar, but also took into account bond financing costs and a 1.75X coverage ratio on senior debt, making its test of financial feasibility more stringent than those used in the other two studies. Yet it still showed greater ability to finance lane construction costs out of toll revenues.
Yet another difference is in the modeling methodologies. All three studiesï¿½Atlanta, the Twin Cities, and Denver--used the existing regional travel demand models, which we all know were not designed to take into account the effect of priced lanes. But each tweaked the models differently, and the Denver study refers to the subsequent use of a micro-model of each corridor, which may account for more of the difference in results.
I draw at least two conclusions from this brief comparative assessment. First, to some extent the transportation planning community has not quite digested the kind of revenue profile which a HOT lane can produce over, say, a 30-year period. So some studies are probably under-estimating the revenue and financing potential. On the other hand, adequately modeling HOT lanes, taking into account their great sensitivity to conditions in the adjacent general purpose lanes, is still at a relatively crude stage. Screening studies of a large number of corridors all at once are probably useful for weeding out corridors with low potential. But their estimates of revenue and extent of construction cost coverage should be taken with large grains of salt. Itï¿½s not until we get to "investment-grade" traffic and revenue studies of specific corridors that we're going to have a realistic idea of the extent to which toll revenues can actually pay for such projects.
Robert W. Poole Jr. is director of transportation studies and founder of the Reason Foundation.