Rudolph Penner: Dynamic Scoring Won't Be Perfect But it is Worth Doing

Rudolph Penner was the director of the Congressional Budget Office from 1983 to 1987, and he is an Institute Fellow at the Urban Institute and a member of the Committee for a Responsible Federal Budget. He wrote a guest post that appeared in the Tax Policy Center blog. It is reposted below.

It is obvious that changes in spending and tax policies affect macroeconomic variables, such as the Gross Domestic Product. The problem is in knowing how much. Different economic models yield very different answers and even within one model, a single revised assumption can change results significantly.

There is also much uncertainty inherent in static analysis. The Congress often considers initiatives for which there is no precedent and analysts may have little data on which to base their conclusions. But there is an important difference between dynamic and static analysis. Dynamic analysis always produces a wide range of plausible results. Static estimates are often much less uncertain.

Nevertheless, the Congress should get the benefit of dynamic analysis. Senator Rob Portman (R-OH) has offered a sensible approach. He’d require a dynamic analysis of significant policy initiatives, but the estimate would not be part of the “official” score.

In doing dynamic scoring for informational purposes only, the Congressional Budget Office or the Joint Committee on Taxation could provide a range of estimates and explain why the results differ depending on the model and the assumptions they use. Individual legislators could decide which analysis they like best, much as a jury must decide which expert witnesses it finds most credible.

The House takes a very different approach. Although CBO or JCT could still provide a range of estimates for informational purposes, they’d be required to prepare a single official dynamic budget score. This would draw attention away from how uncertain such a projection would be.

Dynamic scoring has two troublesome components, although not so troublesome as to justify ignoring dynamic effects altogether.

The first results from the United States being on an unsustainable fiscal path. If no policies are changed, we are headed for a sovereign debt crisis. In such a crisis, stock and bond markets would collapse and government would have to raise taxes and cut spending dramatically. Hyperinflation could also solve budget problems, but only by imposing severe pain on the populace.

Economic actors ought to be aware that such traumatic events would have devastating effects on them or their descendants. If they are paying attention, they should be disturbed by any policy that increases the deficit, because it presumably hastens the advent of such a crisis.

A financial collapse might be prevented or delayed by some future tax increases or spending cuts. Different policy combinations will affect different individuals differently. But this creates a problem for CBO and JCT since they are forbidden from forecasting Congress’ future policy actions.

In a paper that describes its approach to dynamic scoring, CBO says it will assume that private saving rises in response to a deficit increase, presumably because economic actors want to prepare for hard times. But unless the expected policies are specified or the nature of the crisis described, this is not very satisfactory. There is, in fact, no good solution to this problem.

A second problem will arise when Congress enacts a policy that changes a macroeconomic variable by, say, increasing employment. Then it would seem logical for CBO to revise its economic forecast. If the forecast changes, then the baseline for most policies should be adjusted as well, because a change in GDP, the CPI, or the unemployment rate will alter tax revenues and the costs of spending programs.

But imagine a conference committee bargaining over some welfare policy. They are near a conclusion when CBO intrudes and says, “Sorry. All your numbers have changed.” That would drive Congress crazy.

CBO used to provide one baseline early in the year and then change it when they updated their economic forecast in August. The Congress thought that this was too disruptive and after it passed the Gramm-Rudman-Hollings budget law in 1985, it ordered CBO to keep the same baseline all year, even though it became inconsistent with the economic forecast after August. That is to say, the Congress is already behaving illogically. Not changing the whole baseline in response to dynamic analysis adds another layer of illogic, but it probably does little additional harm.

Clearly, any approach to scoring – dynamic or static – is flawed. Does that mean it is not worth doing? Not at all. The most valuable contribution of scoring occurs when it shows that a policy expected to cost very little the first year becomes very expensive in future years. The numbers may be somewhat inaccurate, but the time pattern of costs is unlikely to be far off. It would be unwise to give up on dynamic scoring just because we can’t get it quite right.

It is obvious that changes in spending and tax policies affect macroeconomic variables, such as the Gross Domestic Product. The problem is in knowing how much. Different economic models yield very different answers and even within one model, a single revised assumption can change results significantly.

There is also much uncertainty inherent in static analysis. The Congress often considers initiatives for which there is no precedent and analysts may have little data on which to base their conclusions. But there is an important difference between dynamic and static analysis. Dynamic analysis always produces a wide range of plausible results. Static estimates are often much less uncertain.

Nevertheless, the Congress should get the benefit of dynamic analysis. Senator Rob Portman (R-OH) has offered a sensible approach. He’d require a dynamic analysis of significant policy initiatives, but the estimate would not be part of the “official” score.

In doing dynamic scoring for informational purposes only, the Congressional Budget Office or the Joint Committee on Taxation could provide a range of estimates and explain why the results differ depending on the model and the assumptions they use. Individual legislators could decide which analysis they like best, much as a jury must decide which expert witnesses it finds most credible.

The House takes a very different approach. Although CBO or JCT could still provide a range of estimates for informational purposes, they’d be required to prepare a single official dynamic budget score. This would draw attention away from how uncertain such a projection would be.

Dynamic scoring has two troublesome components, although not so troublesome as to justify ignoring dynamic effects altogether.

The first results from the United States being on an unsustainable fiscal path. If no policies are changed, we are headed for a sovereign debt crisis. In such a crisis, stock and bond markets would collapse and government would have to raise taxes and cut spending dramatically. Hyperinflation could also solve budget problems, but only by imposing severe pain on the populace.

Economic actors ought to be aware that such traumatic events would have devastating effects on them or their descendants. If they are paying attention, they should be disturbed by any policy that increases the deficit, because it presumably hastens the advent of such a crisis.

A financial collapse might be prevented or delayed by some future tax increases or spending cuts. Different policy combinations will affect different individuals differently. But this creates a problem for CBO and JTC since they are forbidden from forecasting Congress’ future policy actions.

In a paper that describes its approach to dynamic scoring, CBO says it will assume that private saving rises in response to a deficit increase, presumably because economic actors want to prepare for hard times. But unless the expected policies are specified or the nature of the crisis described, this is not very satisfactory. There is, in fact, no good solution to this problem.

A second problem will arise when Congress enacts a policy that changes a macroeconomic variable by, say, increasing employment. Then it would seem logical for CBO to revise its economic forecast. If the forecast changes, then the baseline for most policies should be adjusted as well, because a change in GDP, the CPI, or the unemployment rate will alter tax revenues and the costs of spending programs.

But imagine a conference committee bargaining over some welfare policy. They are near a conclusion when CBO intrudes and says, “Sorry. All your numbers have changed.” That would drive Congress crazy.

CBO used to provide one baseline early in the year and then change it when they updated their economic forecast in August. The Congress thought that this was too disruptive and after it passed the Gramm-Rudman-Hollings budget law in 1985, it ordered CBO to keep the same baseline all year, even though it became inconsistent with the economic forecast after August. That is to say, the Congress is already behaving illogically. Not changing the whole baseline in response to dynamic analysis adds another layer of illogic, but it probably does little additional harm.

Clearly, any approach to scoring – dynamic or static – is flawed. Does that mean it is not worth doing? Not at all. The most valuable contribution of scoring occurs when it shows that a policy expected to cost very little the first year becomes very expensive in future years. The numbers may be somewhat inaccurate, but the time pattern of costs is unlikely to be far off. It would be unwise to give up on dynamic scoring just because we can’t get it quite right.

- See more at: https://taxvox.taxpolicycenter.org/2015/02/20/dynamic-scoring-forum-dynam...

 "My Views" are works published by members of the Committee for a Responsible Federal Budget, but they do not necessarily reflect the views of all members of the committee.