Comments on the Penn Wharton Budget Model and Candidate Tax Plans

October 17, 2016

Today the Penn Wharton budget model project presented its macroeconomic analysis of the tax plans proposed by presidential candidates Hillary Clinton and Donald Trump. An interactive version of its macroeconomic analysis can be found here. We expect these results, or similar ones, will soon be incorporated into a Tax Policy Center analysis, though at this time, none has been released.

Under the Penn Wharton budget model’s default assumptions, the Clinton plan is initially a drag on growth during the budget window, but ultimately it increases growth. In contrast, the Trump plan is a boost to growth in early years, followed by a longer-run reduction in growth that begins to materialize towards the end of the ten-year budget window.

Essentially, in both cases, a plan has an initial effect, which then reverses course. Tax Foundation’s modeling of the same plans shows smaller short-run effects, but no course reversal. This post will examine the Penn Wharton budget model and discuss where the results come from, and why Tax Foundation’s numbers come out differently in the end.

The Penn Wharton Model is a Robust, Intelligent Framework for Tax Policy Discussion

Looking at the Penn Wharton budget model as a whole, it is serious work. There are a number of different ways mainstream economists think taxes affect the economy. The Penn Wharton model has the capability to handle essentially all of these effects.

Here is a list of three major effects it has in common with the Tax Foundation Taxes and Growth model:

1. It considers the effects of tax rates on labor supply; the more that workers get to keep the fruits of their labor (rather than paying them in taxes) the more they are likely to work.

2. It also considers the effects of tax rates on savings: the fewer additional taxes you pay on your saving, the more you are likely to save.

3. It considers the effects of tax rates on investment. Firms essentially have two choices on what to do with their cash. They can return their money to shareholders, or they can reinvest. They are likely to have tax liabilities in both cases, and the firm’s decision can be changed as a result of the tax liabilities it would have in each case.

Further, it has two additional effects that don’t feature in the Taxes and Growth model:

4. It considers the effects of budget deficits on what economists call aggregate demand. Aggregate demand is essentially about the exchange of money for goods and services. Higher aggregate demand means there’s more money moving around the economy looking for goods and services to purchase. A budget deficit increases demand, which may result in new people being hired.

5. It considers the effects of budget deficits on the supply of saving available to entrepreneurs and businesses. This is often known as “crowding out.” If you imagine that governments need to find people to lend to them, but so do entrepreneurs, there might be a problem where government deficits soak up some of the saving that would otherwise be used to lend to good private-sector ventures.

These are essentially all five of the main ideas that economists have about tax policy, at the big-picture macroeconomic level. The Penn Wharton model includes all of these, and further, its interactive tools have sliders allowing you to adjust parameters for the model, in case you think some of these responses are smaller or larger than the default assumptions made.

This makes it a great framework in which to discuss things; in a sense, all mainstream macroeconomists are living somewhere in the Penn Wharton model, and the question is mostly how you adjust its parameters.

Tax Foundation does not model the fourth and fifth effects described above. It’s not because we think these theses are fundamentally flawed, but rather, because we think those parameters are relatively small. We’ll discuss each of these in turn.

Why Penn Wharton Assumes Tax Cuts Boost Growth in the Short Run

One feature of the Penn Wharton model is that tax cuts increase aggregate demand, which results in more hiring in the short run. Here is how that effect works: when you cut taxes, the consumers in the Penn Wharton world have more disposable income. Some of them use that extra disposable income by attempting to purchase goods and services. There’s more money bouncing around in the economy after a tax cut. Some of it gets spent on bidding up the prices of existing goods and services (for example, rents might go up on apartments) but some of it also gets spent on hiring new workers who otherwise might not be working. The latter constitutes real economic growth.

The Penn Wharton projection for the Trump plan shows this effect, in the form of speedy GDP growth in 2017. Under their baseline case, GDP grows at 1.6%, but under the Trump plan, it grows at 2.9% during the first year. This is a large effect, and in fact, a much larger effect than Tax Foundation shows for the same year. Penn Wharton is showing the results of the extra tax cut money bouncing around the economy and demanding new goods and services.

Similarly, it shows a drag on growth from the Clinton plan in 2017, with lower GDP than under the current policy baseline. In that case, it is showing the result of less money bouncing around the economy to demand new goods and services.

These are reasonable chains of cause and effect; however, Tax Foundation does not include them, for reasons expressed below.

Why Effects of Tax Cuts on Aggregate Demand May Be Small

Consider a situation with high aggregate demand: perhaps, a large tax cut, like the Trump plan, which results in lots of new money moving around the economy finding goods and services to purchase. A commonly-used metaphor for that situation is that the economy is “heating up.” In some cases, heating up the economy is a good thing; for example, the past eight years have been a time where the economy could use a little more heat. But consistently heating up the economy isn’t a good thing. If more and more money is bouncing around in your economy, year after year, eventually all of that money goes to bidding up the prices of existing goods and services, and none of that results in new hiring. That’s not helpful to anyone, and at that point, you want to cool the economy down.

The Federal Reserve essentially functions as the thermostat in our economy. Its economists monitor when things are getting too hot or too cold, and they manage the currency to try to keep us at an even temperature. If things are heating up, they will raise interest rates and try to pull some money out of the economy. If things are cooling down, they will do the opposite.

So here’s how monetary policy ties into tax policy: think about turning on a heat lamp in a room where the thermostat is set to room temperature. The heat lamp has an effect, yes, but the heat lamp is countered by the thermostat. We see Donald Trump’s tax cut as being something like this. Upon seeing the economy heat up, the Federal Reserve will likely respond by moving to a more contractionary policy than it otherwise would have. We think the Fed carefully monitors economic signals and responds to them relatively quickly. Some budget modelers call this an “active Fed” scenario. But what it means in practice is that tax cuts won’t get the kind of demand-side boost that demand-side fiscal policy modelers predict.

We are open to the idea that the Fed isn’t 100% active, but we do think it is the best starting point for discussion. In other words, the demand-side boost in the Penn Wharton model for the Trump tax cuts is probably too large; in other words, it overstates the short run gains to be had from tax cuts. Similarly, we think the demand-side drag on the Clinton plan looks overstated in the short run.

Why Penn Wharton Draws a Long-Run Relationship between Deficits and Growth

A critical feature of the Penn Wharton Budget Model is constrained saving, in which government and private industry compete for loans from scarce savers. Here’s what that looks like.

For purposes of this discussion, some math might be useful, starting with the basic equation for GDP taught in first-year macro, relating income or GDP (Y) to consumption (C), investment (I), government spending (G), exports (X), and imports (IM).

Y = C + I + G + X – IM

This is just an accounting identity: what it means is that our production is equal to all the purchases we make, plus an adjustment for net trade. We don’t purchase our exports, but we made them, so those are added. Also, while we do purchase our imports, we don’t produce them. So those are subtracted. In the end, you get a figure for GDP (Y) which is uncontroversial and true by definition.

Here’s something else that’s true by definition, relating private saving (S), taxes (T), and consumption (C).

Y = S + T + C

All this says is that for every dollar of income, you either save it, pay it in taxes, or spend it. Simple enough. One thing we can do algebraically is put these two equations together, because both are expressions for total income:

C + I + G + X – IM = S + T + C

Then the expressions for consumption on each side drop out, and we can rearrange the rest of the equation using algebra:

I + G + X – IM = S + T

(G – T) + I = (IM – X) + S

This equation is absolutely 100% true; it’s not a “model” of the economy that makes assumptions, it’s simply the way things are, by definition.

It also nicely illustrates the basics of the Penn Wharton view. On the left side of this equation, we have the borrowers. The deficit is expressed as (G-T), and that deficit requires borrowing from savers. But private investment also requires some sort of borrowing from savers, through either debt or equity instruments.

On the right side of the equation, we have the savers. Domestic saving (S) is just there by definition, but the equation also includes foreign savers. The way foreigners finance things in the U.S. is through trade. For example, they might send us televisions in exchange for dollars, which then get invested in U.S. financial instruments, such as government debt or shares in U.S. corporations. This is reflected in the term (IM – X).

Penn Wharton assumes that for every new dollar borrowed in the U.S., 60 cents must come from U.S. savers, or S, and only 40 cents can come from (IM – X). In contrast, in the Tax Foundation model, there is no such restriction. Foreigners could create the entire increase from the right-hand side of the equation. The reasons for our less restricted view are described below.

Why the Long-Run Relationship between Deficits and Growth is Overstated

Economists all agree that both budget deficits and private borrowing require saving. But the question is how scarce that saving is. At Tax Foundation, we believe saving is not nearly as scarce as the Penn Wharton model shows.

We believe this for a number of reasons: for one, many mainstream economists are discussing ideas like Secular Stagnation and the Global Savings Glut. These ideas suggest there’s oodles of cash on the right side of the equation, looking around for investments to finance and coming up short. As a result, interest rates are coming down in developed countries around the world as the savers bid up the prices of financial instruments.

Under these kinds of circumstances, worries about unavailable saving seem misplaced.

In any event, the particular reasoning behind Penn Wharton’s default assumption (60% closed, as described above) is weak. The assumption was made by looking at who has financed U.S. borrowing on average through recent historical experience. It has been majority domestic savers, hence, the 60% assumption.

But there is no reason to assume that would continue to be the case if there were large changes in the economy (such as a tax cut) that promoted more domestic borrowing but not more domestic saving. Instead, the marginal dollars of new borrowing, under this new policy, could draw different financing than the average dollar of new borrowing under current policy.

The inverse version of our critique also applies. If the government reduced its budget deficits substantially, we would expect a larger fraction of U.S. saving to go abroad, reducing the trade deficit and accumulating foreign financial assets.

Applying this reasoning to Penn Wharton’s analysis of the candidate plans, we think the deficit-induced drag on investment from the Trump plan and the government savings-induced expansion of investment from the Clinton plan over the long run are overstated.

As an additional note, if one believes government savings would induce expansion of investment, this would not happen in the specific case of the Clinton plan, because overall the Clinton tax plan is designed to pay for new spending, not to pay down the debt. Therefore, regardless of modeling assumptions, the long-run expansion projected for the Clinton plan likely would not apply in a full taxes-and-spending analysis of her fiscal policies.


The Penn Wharton Budget Model is a robust and serious contribution to the U.S. fiscal policy debate. It contains the ability to model most of the major effects that mainstream economists discuss as plausible impacts of tax policy. In this post I have identified five major ways that taxes have an impact on economic growth in the Penn Wharton model.

Three of them are similar to effects modeled by Tax Foundation, and involve the simple incentives that taxes have on firms and workers in the economy. On these, we largely agree. For example, Wharton professor Kent Smetters mentioned today that enacting full expensing on new investment would boost long-run investment under the Penn Wharton model, even more so than a corporate income tax cut. The Taxes and Growth model shows the same.

However, two other assumptions warrant further examination. We are skeptical of the assumption that tax cuts will generate new aggregate demand without being offset by monetary policy, though this depends largely on predictions about Federal Reserve behavior. We are also skeptical of the assumption that the majority of a change in desired borrowing would have to be funded by domestic savers; instead, we think foreign savers, through the balance of trade, could easily lend more assets to the U.S. if necessary.

We believe the Penn Wharton model is a great framework, but it would be improved if it assumed a substantially more open economy and smaller demand-side responses. Under such assumptions, the Clinton plan would look better in the short run and worse in the long run than it does under Penn Wharton’s current default assumptions. Conversely, the Trump plan would look worse in the short run and better in the long run than it does under current default assumptions.

In any event, the model introduces a great deal of intellectual substance to public policy debates and should be lauded.



Related Articles