Stimulating fiscal policy and the multiplier

Chapter 12 introduces the idea that relatively small shocks to aggregate demand can be “scaled up” to larger spending changes via multiplier effects. For example, an increase in government spending G of $50 billion can ultimately raise aggregate demand by $100 billion or more. How?

Suppose that the initial $50 billion is spent on building roads. This means that employees of road construction firms receive $50 billion in income. Assuming that federal, state, and local taxes combined imply a typical marginal tax rate of 25 percent, the $50 billion in gross receipts translates into $37.5 billion in income after tax. Let’s assume two-thirds of this is spent. That implies another round of $25 billion in more spending. These secondary expenditures could be for food, or clothing, or recreation, or any number of goods and services. But whatever the spending is for, that $25 billion of extra spending will in turn imply a further $25 billion in income for those employed in providing these goods. If we imagine that they too pay 25 percent ($6.25 billion) in taxes and also spend two-thirds of the remaining $18.75 billion, then there’s a third spending round of $12.5 billion. That’s income for someone else leading to further spending, and so on. At the end of the day, the total cumulative effect of the initial $50 billion rise in government spending is:

This analysis helps explain why the effectiveness of a fiscal stimulus is often measured by the size of the associated multiplier. In the example above, we say that the government spending multiplier is 2, because the rise in G of $50 billion ultimately led to a rise in aggregate demand and income that was twice as big. Note, too, that if this were truly the case, it would imply that the $50 billion rise in public sector production was matched by a subsequent equal rise of $50 billion in private production. That would generally be considered a very effective stimulus.

In contrast, if the multiplier had simply been 1, that would imply a weaker but still effective stimulus in which the rise in public sector spending simply led to an equal rise in the production of public goods — public highway construction in the example. Had the multiplier been less than 1, this would mean that the rise in public sector output had been partially offset by a fall in private sector production. And had the multiplier been zero, the spending stimulus would have been totally ineffective. All it would have done is shift $50 billion worth of production from the private sector to the public sector with no net gain at all.

We can also measure the multiplier for a tax cut. Thus, if a tax reduction of $50 billion ultimately causes real GDP to rise by $75 billion, the tax cut multiplier is 1.5. But if it causes real GDP to rise only by $16.67 billion the multiplier falls to 0.333, in which case we would have to say that the tax cut was fairly, though not totally, ineffective.

The examples we’ve been talking about so far — public spending hikes or tax cuts — are examples of expansionary fiscal policy. The multipliers for the opposite case of contractionary fiscal policies, such as spending cuts and tax hikes, are in principle the same but, of course, depress demand and real GDP.

The effectiveness of any fiscal policy is reflected in the associated multiplier. A spending rise or tax cut of $X that leads to a rise in income of $μX has a multiplier of μ. Similarly, a spending cut or tax increase of $X that ends up reducing real GDP by $μX also has a multiplier of μ. The larger μ is, the more effective is the fiscal policy. A multiplier value of μ = 0 implies a totally ineffective fiscal change.

Context is important here. Starting from a position of full employment with real GDP equal to potential and the room to increase GDP limited at best, the multiplier for any expansionary fiscal policy, whether it’s a spending rise or a tax cut, is likely to be quite small. Why? Remember, Y = C + I + G + NX. In this setting, with Y unable to rise much, an increase in government spending G would have to be offset by a decrease in either consumer spending C, business investment spending I, or net exports NX (or all three combined). Likewise a tax cut that raised household disposable income Y – T, by lowering T and so induced a rise in consumer spending C, would have to be offset by a fall in either investment I or net exports NX, assuming that the political authorities keep government spending G unchanged.

In this light, it’s important to remember the usual setting for a fiscal stimulus. It’s typically a case in which, as in Figure 14-4 in the previous chapter, there has been some shock that’s lowered aggregate demand and therefore shifted the AD curve inward. The situation is shown again here in Figure 15-1. A negative demand shock has shifted the AD curve from AD₁ to AD₂. As a result, real GDP has fallen from the full-employment potential level of Y₁ = Y* to Y₂ < Y* as the economy moves down and to the left along the short-run aggregate supply curve SRAS₁. The short-run equilibrium has moved from point A to point B.

We noted in Chapter 14 that at point B, policy-makers have basically to choose one of two options. One is to wait for the long-run adjustment in which wages and prices gradually decline, implying that the short-run supply curve moves to SRAS₂ and a full-employment equilibrium is re-established at C. The other is to try to use the tools of macroeconomic policy to push the aggregate demand curve back to AD₁. In principle, this could be accomplished by either monetary or fiscal policy. In this chapter, though, we are focusing on fiscal policy. (Yes, in reality the two policy teams typically coordinate. It’s easier though to examine each separately, which is what we’re doing here.)

The question then is: How much can changes in government spending G or taxes T move the aggregate demand curve and raise real GDP? More directly, what are the spending and tax multipliers at a point like B? These are the questions at the heart of the debate over the Obama stimulus.

Doubting the multiplier effect: Fiscal skeptics

Let’s start with the critics. It turns out that even in a recession, there are reasons to suspect that the multiplier for any fiscal policy is very small or zero. Broadly speaking, there are three reasons why this may be the case. The first, oddly enough, has to do with interest rates and the demand for money. The second reflects doubts about the link between current income and current consumer spending. The third has to do with the government budget constraint we talked about in Chapter 13 (check out the “Acknowledging the government budget constraint” section there).

Crowding out, money markets, and interest rates

As discussed in Chapter 14, the demand for money expressed in purchasing power terms M / P = m depends negatively on the interest rate but positively on the level of real GDP Y. Figure 15-2 shows this relationship.

In the figure, the interest rate effect is reflected in the negative slope of each money demand curve, LL and L'L'. For instance, along demand curve LL associated with income Y₁, a rise in the interest rate from i₁ to i₂ reduces money demand from m₁ to m₂.

However, if income rises from Y₁ to Y₂, the demand for money curve shifts from LL to L'L'. This means that if the money supply stays at m₁ and the interest rate stays at i₁ there will be excess money demand. To choke this demand off, the interest rate would have to rise to i₂.

Now fiscal policy is about government spending, taxes, and, if necessary, issuing debt. It’s not about printing more money. Therefore, when we consider tax cuts or spending increases, we hold the money stock constant at some level, such as m₁ earlier.

Perhaps you can begin to see one of the sources of doubt about the size of the fiscal multiplier. To raise real GDP, any expansionary fiscal policy has to shift the AD out. But as soon as a rise in government spending G or a cut in taxes T starts to do this — as soon as income starts to rise — it also begins to shift the money demand curve outward and so starts to raise the interest rate as well, just as we saw in the shift from LL to L'L' in Figure 15-2.

In turn, because consumer spending C and investment spending I are both sensitive to interest rates, both fall, and this tends to offset the original fiscal stimulus. The increase in G, in other words, “crowds out” spending for C and I. If that’s not enough, the rise in the domestic interest rate will make domestic assets more attractive to foreign investors. As they try to buy U.S. assets, the value of the dollar will rise, and this will reduce net exports NX.

How much interest rates will rise and how much this crowds out consumer and business spending, C and I, or reduces net exports is a matter of some question. The evidence suggests that the effect is there but that it is not big enough to undo the stimulus effect totally. We’ll come back to this point in a little bit.

Consuming now or consuming later

It’s the first of the month, and you’ve just received your monthly paycheck of $6,000. It’s also Saturday and the night will soon be here. You’re starting to make plans. Take a date to dinner? Join friends at the movies? Do a pub crawl?

With so many good choices, it’s hard to know which to choose. There’s one thing that we can be reasonably sure of, though—you won’t spend all $6,000 this first night (unless you do the pub crawl and lose your mind). If you do, you’ll have to go the next four weeks without any cash and life will be pretty Spartan. So, when you get your Saturday pay, you’ll smooth out your spending fairly evenly over the next 30 days or so—until another paycheck comes and you can start over again. As a first approximation, you’ll spend about $200 each day.

This same intuition applies over longer time spans. In choosing how much to consume today, people realize that they’re also choosing how much they’ll consume in the future. And the goal is to keep consumption relatively smooth — to avoid binges in one year when times are good that then force you to live like a hermit for several years after.

Thinking about consumption decisions this way seriously weakens the link between your income in any period and your consumption in that same period. Go back to the payday example. In the first day you received income of $6,000 and spent only one-thirtieth of it, or $200. And on each of the next 29 days, you again spent $200 even though you had no income at all. That’s because you saved $5,800 from your first-day income and that nest egg — that wealth — allowed you to spend and enjoy life over the following two weeks even though you received no additional pay. That’s really a way of saying that it’s your wealth that truly matters for your consumption spending. That’s why the elderly can enjoy retirement. Although their income is low, they’ve spent their working years accumulating a nest egg for that very purpose.

Once again, we can apply this approach when the main interval of time is a year and not a day. When we do, we can see that a fiscal stimulus that changes your current income but not your wealth will not change your spending very much.

For example, suppose the government gives everyone a $6,000 tax rebate for just this year. The analysis under discussion says that if a typical person has 30 years or so to live, they’ll only spend $200 of the rebate in any given year. So, the short-run spending impact of the tax cut is pretty weak.

A rise in government spending will be a little more powerful — at least there we get the first round effect. Go back to the road-building example. The government’s $50 billion on new construction will raise total spending by that amount right away. But again, if the typical construction worker is looking at a spending horizon of 30 years or so, there won’t be much more spending beyond that initial $50 billion. They’ll save the bulk of it to spread evenly over the years ahead.

In short, the multiplier won’t be much more than 1 for government expenditures and will be a lot lower for tax cuts. And all this leaves out the interest rate effects from our earlier “crowding out” analysis. If we add these into the mix, the likely multiplier estimates shrink even more.

Taxing now or later: The government budget constraint

Time for a riddle: When is a tax cut not a tax cut? Answer: Always — at least according to some economists. Why? Because of the government budget constraint we discuss in Chapter 13. (Have a look at the “Acknowledging the government budget constraint” section there.)

Remember, the government ultimately has to pay off its loans just like anybody else. A tax cut today means more borrowing today. And that means the government will have to raise taxes more in the future. A simple bit of arithmetic shows the basic idea.

Suppose we divide the world into two periods: the present (period 1) and the future (period 2). If it helps, you can imagine that each period is long, say 20 years or so. We’ll imagine that the budget is initially balanced in both periods. T₁ = G₁ and T₂ = G₂. If that’s the case, then if the government cuts taxes by the amount ΔT₁ in period 1, it will run a deficit of exactly that amount. So, it’ll have to borrow B₁ = ΔT₁

In period 2, the deficit — principal plus interest — comes due. As this is the last period, the government must pay off its first-period debt principal plus interest. Because period 2 taxes were originally just enough to cover period 2 spending, this means that period 2 taxes have to be raised by the amount ΔT₂. Specifically, ΔT₂ = (1 + r)B₁.

Now you might say, “Okay, but we still got our first-period tax cut. Doesn’t that change our first-period spending?” Not really. As we saw in the last section, a sensible consumer will think about future spending when she gets her income at the start of each month or year. So, she’ll see the future tax increase as well as the current tax cut. Suppose you knew that you’d be facing a big bill in a few years — maybe to pay for college or some similar expense. Wouldn’t it be a good idea to start saving now?

That’s exactly what fiscal policy skeptics argue. Cutting taxes today means issuing debt today. (Again, if we print money we’re using monetary policy and we’re focusing on fiscal policy alone.) But debt today means more taxes in the future. Sensible (rational) consumers will understand that. As a result, they’ll increase their savings to pay the larger tax bill coming down the line.

How much will savings increase? Let the first-period tax cut be ΔT₁. Then the amount of first-period debt is D₁ = ΔT₁. By the time the second period rolls around, the debt plus interest will be D₁(1 + r). That’s how much taxes will have to rise to pay off the loan. So, ΔT₂ = D₁(1 + r).

Consumers want to raise their first-period savings by an amount ΔS₁ to pay the anticipated tax rise. Of course, they know that whatever they save will also earn interest. So, the amount they need to raise their first-period savings satisfies the condition that ΔS₁(1 + r) = ΔT₂. But ΔT₂ = D₁(1 + r). So, the needed increase in savings is ΔS₁ = D₁ = ΔT₁.

In short, consumers should just save the entire first-period tax cut and put it in the bank. Then they’ll have just enough to pay the extra taxes later. To put it another way, issuing debt to pay for current government spending is really the same as raising taxes now to pay for government spending. All the debt does is change the timing of taxes — you pay more in the future instead of today. It doesn’t really change the amount of taxes you have to pay overall.

Ricardian equivalence is the idea that debt finance and tax finance are really the same. If it’s true, it’s another reason to doubt the efficacy of fiscal policy. A tax cut will just be saved. A similar result holds for an increase in government spending financed by borrowing. That extra debt again means more taxes later. So, again, consumers should save the amount of the spending hike so as to pay off the higher taxes that they know are coming.

Locking interest rates in the liquidity trap

As we saw in Chapter 14 (check out the section “Zeroing in on the liquidity trap”), the demand for money flattens out as interest rates get close to zero and the economy falls into the liquidity trap. In the liquidity trap, though, interest rates are locked at a low level. Even if the demand for money increases due to a rise in income, interest rates won’t rise very much, if at all.

This often happens in a serious recession. When real GDP is seriously low, businesses have no reason to build new plants and so have no need to borrow. Likewise, consumers are wary of taking on more debt if the economic outlook looks shaky. So, the demand for credit and hence the price of credit is low. The case is illustrated in Figure 15-3.

In the figure, the economy is in a recession. The demand for money and credit is low relative to the money supply, and the interest rate is i₀. Expansionary fiscal policy shifts the money demand curve from LL to L'L'. Interest rates do rise but not by very much, to i₁. They won’t rise at all if the economy is on the completely flat part of the money demand function. Of course, had the economy been operating closer to potential GDP, the demand would have been pushing more against the available money supply and we would have been in a higher interest rate region, such as i₂. Here, the shift from LL to L'L' would have had a much larger effect on interest rates. But that’s not the case in a serious downturn. So, crowding out is not an issue, and expansionary fiscal policy can be helpful in that setting.

Moreover, even if interest rates do rise, there is some question about how sensitive consumer C and business investment I spending are to interest rate changes in the short run. Households and firms plan their spending for large items in advance. Even if interest rates rise a fair bit, they may stick with these plans. If the old car is dying, you need a new one whether the finance charge is 2 percent or 4 percent. But if interest rate changes don’t affect spending very much, then again, crowding out is not an issue. Expansionary fiscal policy may lead to higher interest rates, but that doesn’t lead to lower C and I spending.

Constraining liquidity

As we’ve seen, two sources of doubts about fiscal stimulus both rest on the idea that consumers are rational and forward-looking. As a result, they spread any income windfalls over several years into the future and they anticipate the higher future taxes that government deficits today imply.

One simple counterargument is that consumers really aren’t that rational. In this view, they don’t look that far into the future. Hence, their current spending is just based on their current disposable income. They live for today. This means that they also don’t worry about future taxes.

Of course, that may well be true. But it’s an uncomfortable argument for an economist to make because the discipline derives its primary results from models assuming rational behavior. It’s also hard to take it as a general description of how people act. Tons of advertisements for companies that will help you plan financially for retirement or for your children’s college years make clear that many if not all people are thinking many years ahead.

There is, however, a good reason to believe that some households will base their current spending decisions on their current income even though they’re as rational and forward-looking as anyone else. The reason has to do with credit market imperfections or what macroeconomists call liquidity constraints.

Liquidity constraints refer to the fact that some households (and some businesses) have either limited or no access to credit. Hence, their spending is limited to whatever liquid resources they have — cash and current income for the most part. They may (rationally) want to spend more but they cannot.

For example, suppose you’re lucky enough not only to live in the San Francisco Bay Area but also to have a rich uncle who loves you dearly. Sadly, you discover that your uncle who is 96 (and only recently discovered that he likes older women) also has a terminal heart condition that will take his life within the next three years. You know, though, that he plans to leave everything — $9 million — to you.

You’re going to have plenty of cash in the future, but why wait? Why not take out a loan now and buy a two-bedroom condo in San Francisco? Once people understand your situation, they’ll know you’re good for the money and happily lend to you … or not. Actually, you’d probably have a hard time getting a loan, unless your uncle came and co-signed the agreement. On your own, it will be difficult to persuade a bank loan officer to agree to a condo mortgage — or any other major loan.

You won’t be alone either. Many households and businesses find that even though they’re confident of having substantial resources later, they can’t borrow against that future easily. Medical and law students — even those near graduation — often have difficulty getting big loans. College graduates fresh to the workforce frequently do as well, even if they have good starting salaries. Small firms can find themselves locked out of the loan market despite the fact that their business is sound.

And even when many can get credit, they can only borrow at a very high interest rate. You can put your money in the bank and earn, say, 1 percent. But when you want the bank to put some money into you via a mortgage or personal loan, you may pay 5 percent or more.

Whenever someone who wants to borrow simply cannot get funds or can do so only at a significant premium over the interest rate they can earn, their spending becomes constrained by their own liquid resources. This has two, closely related implications: First, even though such households and firms may look far into the future, their spending today will tend to fall and rise with their current income. Suppose you have an income of $40,000 this year but $60,000 next year. Ideally, you’d like to avoid sharp swings in your consumption and consume roughly even amounts of $50,000 in each period. But this requires that you borrow today $10,000 more than you actually have. If you can’t do that, the best you can do — the smoothest you can make your consumption stream — is to consume all $40,000 this year and all $60,000 the next.

Further, if your income falls by $5,000, then so does your current spending if you can’t borrow. The same holds true if it rises by $5,000. In that case, the closest you can get to having the same consumption in each period is to spend all of the extra $5,000 now. Then your consumption sequence over time goes from first $40,000 and then $60,000 to first $45,000 and then $60,000.

So, returning to our first expansionary fiscal policy example, the construction workers who are paid $50 million will spend a lot of it if they’re liquidity constrained. And so will those who receive the payments in subsequent spending rounds. Likewise, the taxpayers who get a tax cut will do the same thing. That brings us to the second implication of liquidity constraints.

Think back to your rich uncle. In anticipation of your inheritance, you wanted to borrow from the bank today but couldn’t. Yet what if the government comes along and cuts your taxes? Yes, that means the government is borrowing, and now you’ll have to pay more taxes in the future. But you also would have had to pay more money back if you had got the loan you wanted and not the tax cut. What’s really happened, then, is that the government has done the borrowing for you. It’s as if the government went to the bank, borrowed the money, and gave it to you. Some time down the road, they’ll ask you for the funds to help pay that debt off. However, that’s what you wanted in the first place. In other words, the government’s tax cut has made it possible to do the spending today that you wanted to do but couldn’t. And that’s what you’ll do. You’ll spend the extra disposable income instead of saving it, as predicted by Ricardian Equivalence.

Rationing liquidity and adverse selection

It’s worth noting that liquidity constraints and tight credit can happen even when interest rates are relatively low. To see why, you have to realize that banks and other credit institutions can’t always use the price of credit — the interest rate — to allocate credit. Some non–price rationing is likely to be a necessity.

Imagine you’re a loan officer at a small bank, and there are two potential borrowers, each of which wants to borrow the $100,000 you have available to lend. However, SafeCo has a project that in one year will be worth either $100,000 or $106,000, each with a 50 percent probability. The other firm, RiskyBusiness, has a project that will be worth either $88,000 or $112,000 in one year, again with equal probability. You can break even so long as you earn 2.9 percent, but seeing that there’s excess demand, you start by charging 5.99 percent. A borrower of $100,000 has to pay you $105,990 a year from now or declare bankruptcy. What will happen?

SafeCo will work out that half the time, its project fails and is worth only $100,000. In that case, they declare bankruptcy and let you keep the $100,000 that’s left. The other half of the time, the project will pay $106,000. In that case, SafeCo will pay the $105,990 your bank is owed and then walk away with the $10 in pure profit. On average, SafeCo wins $5 ($10 half the time). So, it will accept your terms. Your bank earns zero return half the time and 5.99 percent the other half, or roughly 3 percent on average — just over the minimum you need — and you get your principal back no matter what.

RiskyBusiness however will be willing to borrow at 5.99 percent, too. If you lend to them and their project goes belly up, they declare bankruptcy and leave you holding a less-than-properly-full bag of $88,000 worth of assets. If its venture pans out, though, they earn $112,000, pay your bank the $105,990 owed, and walk away with $6,010 in profit. On average, RiskyBusiness wins $3,005. However, your bank loses $12,000 half the time and gets nearly $6,000 the other half for an average loss of about $3,000 –a roughly 3 percent return.

Obviously, you’d prefer lending to SafeCo instead of RiskyBusiness. But in the real world, you can’t tell which firm is which. All you know is that at a 5.99 percent interest rate, there are still two firms wanting a loan of $100,000, and that while you can only lend to one of them, the choice is important. On average, you earn about +3 percent on one loan, but –3 percent on the other.

Whenever there’s excess demand, economists often argue that the price — in this case, the interest rate — will be bid up until enough buyers drop out. Hence, you consider raising the interest rate to, say, 7 percent. But a little further thinking makes you stop. Because at an interest rate of 7 percent, SafeCo will be the one dropping out. At that rate, it can’t make a profit even in the best of times. You’ll have succeeded in getting rid of the excess demand for your funds. But the borrower that’s left will be the one you don’t want to lend to.

The problem is that in many markets, prices not only equate supply and demand but also sort — they affect the kind of producer or consumer that participates. In some cases, this sorting effect is benign, but in a number of cases, it sorts or selects adversely. If there are two kinds of used cars—lemons that are worthless and reliables that are worth $10,000 — and if some seller accepts your offer of $5,000, then be afraid … be very afraid. The owner of a used car likely knows exactly what kind he has, and only a lemon-owner will sell at that price.

Adverse selection refers to a common problem in many markets in which prices not only equate supply and demand but also select the quality of the buyer or seller that participates. In a number of cases, this selection is adverse in that the poorest quality buyers or sellers are the ones selected.

Insurance markets are a classic example of adverse selection. Suppose there are equal numbers of two types of drivers: crazies and nerds. Crazies have a 20 percent chance of a $10,000 accident, and nerds have only a 10 percent chance. So, a fair insurance policy premium for the whole population would be $1,500. If a random group of 100 drivers all insured at this price, 20 percent of half of them — 10 drivers — would have a $10,000 accident; and 10 percent of the other half — 5 drivers — would, too. So, the total damages would be 15 × $10,000 = $150,000. With all 100 drivers paying $1,500, that’s exactly how much the insurance companies would take in. They’d break even (which is what competition does). All drivers would be insured, and life would be good.

Because of adverse selection though, that would not be an equilibrium. In particular, each nerd is paying a $1,500 premium even though he only has an expected damage amount of $1,000 (0.10 × $10,000). So the nerds will drop out, leaving only the crazies in the market and forcing the insurance companies to charge a premium of at least $2,000 — at which price it stops making sense for even the crazies to insure. So, the whole insurance market breaks down. It’s to avoid this outcome that virtually every state has mandatory driver’s insurance. It’s also the reason that the Affordable Care Act, or Obamacare as it’s typically called, includes a mandate to buy health insurance.

Getting back to the banks, we’ve seen that they can face an adverse selection problem too. If there’s excess demand, raising the interest rate on loans can make things worse because the sorting effect leaves them with a pool of riskier borrowers. So interest rates may stay low — but that doesn’t mean banks will do much lending. Unable to judge the credit-worthiness of potential borrowers, the banks may just sit on their funds. Credit can be tight — few consumers or businesses can get loans, even though interest rates are relatively low.

Sinking into the liquidity trap

As we noted earlier, the Great Recession came on swiftly. In the year leading up to the second quarter of 2009 — when the new president had been in office less than three months — real GDP had fallen by roughly 4 percent. The fall was especially severe in credit-sensitive spending categories. Housing markets had collapsed. Non–residential equipment investment had fallen by 20 percent. Unemployment had surged from 5 percent at the start of 2008 to 8.5 percent at the end of 2009’s first quarter.

The onset of the financial crisis had also impacted the financial markets. Treasury interest rates had collapsed. The rate on 3-month Treasury bills fell from 5 percent at the end of 2007 to just 0.3 percent by February 2009. At the time of this writing, it has not risen above that level since. Likewise, the federal funds rate followed a similar trajectory, falling sharply and essentially hitting zero (or very close to it) for the next 7 years.

This is a case where it’s important to recognize that there is not one but many interest rates. That’s because the subprime crisis and the bankruptcy or near-bankruptcy of some of America’s largest and most well-known firms — think GM, Chrysler, Lehman Brothers, and AIG, among others — created a tremendous unease about lending to anyone other than the government. So, investors rushed into safe U.S. government and very short-term almost cash assets, while they fled from other markets. So other important interest rates initially rose. This includes the Baa corporate bond rate, a key indicator of business credit conditions. It also includes consumer credit card interest rates.

Eventually, Fed policy and the sluggish economy brought all interest rates down. Whereas the 30-year fixed-rate mortgage interest was 6.7 percent in mid-2007, it fell steadily to an all time low of 3.33 percent. It has remained low ever since (about 3.7 percent at this writing). Similarly, the Baa rate has been hovering around the 5 percent mark for the past four years. But it initially rose to 9.2 percent in the midst of the financial crisis. As we’ll see, that initial rise in key rates reflected a new nervousness about making loans.

To many macroeconomists, this data — especially from 2010 on — paints a compelling picture of an economy falling into a liquidity trap, or something close to it. Much like the financial shocks of 1929 and the Great Depression they triggered, the financial crisis of 2007–08 seemed to be bringing the U.S. economy down once more into a sustained period of economic stagnation characterized by persistently high unemployment and low interest rates.

Rationing credit … sharply

The recession came in the immediate wake of the 2007–08 financial crisis. That meant that banks and other institutions that had suffered large losses had to rebuild their capital before they could take on new loans. At the same time, the main lesson of the subprime debacle was that banks needed to be a lot more careful about extending credit. Stricter standards would have to be met before a would-be borrower could get funds. Unfortunately, a lot of those would-be borrowers had also suffered losses. Their assets, and therefore the collateral they could offer in any loan deal, had fallen a lot in value. They’d have had trouble meeting the old standards, let alone these new, tougher ones.

All this suggests the onset of a credit crunch. The first bit of evidence is the sharp difference that emerged early on between the interest rates on safe assets like U.S. Treasuries and more risky ones like Baa-rated corporate bonds. As noted earlier, the size of that gap can be taken as a rough indication of how risky the banks perceived business and consumer loans to be. So, although the 3-month Treasury rate fell quickly to 0.3 percent, the Baa rate rose from 6.5 percent to over 9 percent during 2008. Yes, it slowly came down later along with other rates — but that first big jump reveals that the markets felt that lending was now a lot riskier.

And, as we pointed out earlier, even though the Baa and other rates eventually fell, that doesn’t mean credit became easy. Many homeowners found themselves underwater — meaning the value of their homes had fallen below the balance on their mortgage. Because refinancing typically means paying off the first mortgage and then obtaining a new one, these owners were unable to refinance and take advantage of the now lower mortgage rates. And even those whose houses were still “above water” often had lost enough value that they no longer qualified for home equity loans, especially given tough new standards that surveys of bank loan officers show had been imposed. This was true for commercial and industrial loans as well as for consumer installment loans. Obviously, we can’t observe how many loans were not made. What we can observe is that the only small businesses and consumers getting loans were now those with much higher and more stable incomes.

We can also see that banks began to hold a lot more reserves than required. Virtually zero for the prior 30 years, bank excess reserves shot up sharply to over $1 trillion by the end of 2009. Banks were simply not mobilizing reserves to make loans and extend credit the way they had been.

In short, credit was tight, and it is likely that many households and firms were liquidity-constrained, especially in the early part of the recession. They would likely spend any liquid funds they could get their hands on. So, the economic conditions were probably as favorable for expansionary fiscal policy to work as they could be.

Stimulating success

Macroeconomists generally agree about the long run. But any consensus often disappears in the short run, especially when it comes to policy debates. It can even get a little nasty, with some calling their opponents’ arguments “dog whistles” and throwing words like “nitwit” (and far worse) into the mix.

In the case of the stimulus, one difficulty is that it had many components — construction spending, tax cuts, and block grants to states — and each will have a different multiplier. Another problem is that economists can’t run laboratory experiments. They can’t have one Petri dish with an economy in recession that enacted a stimulus program and compare it with one that had an identical recession but no stimulus. So, in evaluating the stimulus, economists have to somehow work out what the economy would have looked like without it. That’s tricky business.

For example, not all states received the same stimulus spending. You might think we can measure the stimulus impact by looking at, say, unemployment in those states that received a lot of stimulus spending and those that received only a bit. The problem is that one reason a state may have received a lot of stimulus money is precisely because it had a lot of unemployment. If we don’t correct for that, we can easily find a correlation between stimulus spending and high unemployment and conclude that the stimulus was harmful, even though it may have lowered unemployment everywhere.

This is what economists refer to as the endogeneity problem. It’s like finding a correlation between the probability of dying and being in hospital. Yes, a lot of people die at hospitals. But that doesn’t mean hospitals are bad for your health. Although there are a lot of statistical techniques to surmount the endogeneity issue, none is perfect, and macroeconomists very often disagree about which ones are best.

That’s why it’s so surprising that a very large majority of macroeconomists believe the Obama stimulus basically worked. A well-known survey conducted by the University of Chicago’s Booth School of Business asked 55 prominent economists for their views on the effect of stimulus program. Of course, not all agreed. Five did not answer. But of the remaining 50, 44 agreed that the stimulus had worked to raise employment substantially higher than it would have been. Economists are always uneasy about unanimity. That’s about as much consensus as you’ll ever see in macroeconomics.

This general opinion seems buttressed by academic research as well. At least a dozen formal studies of the ARRA’s economic impact have been conducted. About three fourths of these find multipliers that, taken together, range between 0.5 and 2, with the 1.25 midpoint or maybe a bit higher being a not bad summary measure. These studies imply that the stimulus added on the order of 1.5 to 3 million jobs from 2009 to 2011 over what would have happened in the absence of the expansionary fiscal policy. Of course, there is dissent. It wouldn’t be macroeconomics without that. But as the Chicago survey indicates, the dissent is a small minority.

Again, it’s important to note that the conditions were exactly the kind that favored fiscal policy. Output was well below potential, so there was lots of slack in the economy. Financial markets seemed to have fallen into a liquidity trap or something close to it. Hence, the interest rate pressures from any fiscal expansion were minor. And the Fed’s monetary policy insured that things stayed that way. Finally, there was reason to believe that a number of households and businesses were liquidity-constrained and would respond by spending a lot of any income generated by the stimulus.

This is all a way of saying that there’s a warning label on the other side of this experiment: Fiscal policy will not be that effective when these conditions are not met. We talked about the various lags that make monetary policy difficult in Chapter 14. (Take a quick look at the section “Accounting for policy time lags.”) Those same lags of recognizing the problem, deciding what to do, and then waiting for the action to have effect also hamper fiscal policy. The decision lag is especially long here because fiscal changes have to be acted on by Congress … an institution that has never been known for quick, uncomplicated action.

Therefore, it’s probably a good idea to reserve the use of active fiscal stabilization to really clear cases like the U.S. economy in 2009. With small cyclical bumps, the margin for error is smaller, too. Trying to smooth these out by making major changes in taxes or spending could easily make things worse.

And don’t forget the long run. Once taxes are cut and spending programs are launched, it’s often politically hard to undo them. That could lead to persistent deficits, accumulating debt, and a dangerously large debt-to-GDP ratio. (Check out the section “Understanding the burden of the debt” in Chapter 13.) Maintaining a commitment to fiscal prudence in the long run requires a fair bit of fiscal restraint in the short run.

Stabilizing automatically

In any case, there are some fiscal responses that help protect the economy against minor disturbances automatically, without any major legislation. Suppose your landlord said your rent would go down whenever you get laid off. That would certainly soften the blow … and allow you to keep spending a fair bit for food, clothing, entertainment, and other items. Well, when it comes to taxes, the government is like this fairy-tale landlord.

Remember, the government sets tax rates, not actual taxes. If you’re in a 25 percent tax bracket and your gross income falls by $20,000, your disposable income only falls by $15,000. In effect, the government reduces your tax bill by $5,000 when your income falls. In fact, it’s better than that because the tax schedule is progressive. When your income falls by $20,000, you may also move into a lower tax bracket. Instead of being taxed at 25 percent, you may only be taxed at, say, 22 percent on the gross income you continue to earn.

The tax schedule is a way that fiscal policy automatically responds to an economic downturn on the government revenue side. But it also responds on the spending side. As you and others get laid off, the government pays more in unemployment compensation. It also pays out more in food stamps and other social insurance programs.

Automatic stabilizers are taxes and transfers like unemployment compensation that change automatically with the state of the economy. In an economic downturn, the government automatically collects less in taxes and provides more income support through unemployment compensation and other programs. As a result, people who lose their jobs or move to part-time work may still have significant disposable income. Therefore, they can continue to buy things. In turn, this means that aggregate demand doesn’t fall by as much as it would otherwise.

Automatic stabilizers help stop a small recession from turning into a big one. They also mean, of course, that the deficit widens automatically if the economy begins to slow. The U.S. Congressional Budget Office (CBO) has estimated that for every $100 that real GDP drops below potential, the deficit grows by $35. So, over one quarter of the fiscal year 2015 federal deficit was due to the economy still operating below potential.

The automatic stabilizers are one reason most economists oppose requiring the budget to be balanced each year. Suppose the economy is humming along nicely with a balanced budget and real GDP at potential. For some reason, it suddenly begins to slow. Real GDP starts to lag, and unemployment begins to rise. As a result, tax collections fall off, and unemployment compensation payments start to rise. If the budget were required to always be in balance, the government would be forced to adopt contractionary fiscal policy — either raise taxes or cut spending or both. That would make the recession worse.

We may not want to adopt expansionary fiscal policy every time the economy stumbles. But there’s no reason to shoot ourselves in the foot either.