Changes in fiscal policy have been substantial over the past two years, including passage of the Budget Control Act of 2011, which led to sequestration spending cuts. In addition, temporary payroll tax cuts expired and income tax rates for higher-income taxpayers rose following passage of the American Taxpayer Relief Act of 2012. Two important questions are how much has federal fiscal policy been a drag on growth in the recovery to date and to what extent will it affect growth over the next few years? Moreover, is this fiscal drag unusual or part of the normal pattern in which government spending tends to fall and tax collections tend to rise as economic activity gains momentum?

In this Economic Letter, we examine these questions by estimating what fiscal policy would be if it followed historical patterns in the relationship between fiscal policy and the business cycle. We then compare this historically based estimate with actual fiscal policy during the recession and recovery to date. We also look at government projections of fiscal policy over the next three years to see how these compare with estimates based on the historical norm. Finally, we discuss what these trends in federal fiscal policy imply for economic growth.

The historical norm of federal fiscal policy

Historically, fiscal policy tends to be expansionary in recessions. As economic activity slows, tax revenue falls and government spending rises, giving a boost to the economy. The opposite occurs during expansions, as tax revenue rises and government spending falls. Much of this countercyclical pattern is by design. During a recession, so-called automatic stabilizers kick in. These are programs that boost government spending and reduce tax receipts without explicit legislative action. For example, income taxes fall and unemployment insurance and Medicaid automatically rise during downturns, adding to the federal deficit and ideally stimulating economic activity. During upturns, the process automatically reverses as spending on safety net and income-support programs falls and tax revenue rises, trimming the federal deficit.

Much of the analysis of the countercyclical effects of federal fiscal policy only takes these automatic stabilizer programs into account. For instance, the Congressional Budget Office regularly produces estimates of the cyclical component of the federal deficit based on automatic stabilizers (CBO 2013). Yet, automatic programs are only part of the picture. Discretionary fiscal policy, that is, legislated tax and spending changes, often tends to be countercyclical. Congress commonly passes temporary tax cuts or stimulus spending to counteract downturns. Therefore, to fully capture the cyclical effects of government budgetary trends, we consider both automatic stabilizers and discretionary fiscal policy.

To analyze whether recent fiscal policy has followed historical patterns, we use a statistical model of the relationship between fiscal policy and the business cycle based on three variables: government spending other than interest payments, tax revenue, and the difference between the two, which is known as the primary deficit (see Lucking and Wilson 2012). We measure these over time as a share of gross domestic product. We measure the business cycle using the difference between actual GDP and the CBO’s estimate of potential GDP, which is known as the output gap.

Our model allows us to estimate what fiscal policy is likely to be at any point in time given the state of the business cycle. We call these estimates the historical norm since they are based on the average historical relationship between fiscal policy and the business cycle. We can also use this model to estimate a historical-norm level of fiscal policy in coming years given CBO’s projections of the output gap.

Fiscal policy in the recession and the recovery

Figure 1 compares actual fiscal policy with estimates based on the historical norm for noninterest federal spending, tax revenue, and the primary deficit. It shows that federal fiscal policy was unusually expansionary during the Great Recession. Federal spending grew more and tax receipts fell more than usual, even taking into account the recession’s severe depth and duration, and the resulting very large output gap. This reflects both automatic stabilizers and discretionary changes in spending and tax policy, such as the American Recovery and Reinvestment Act, the economic stimulus program passed by Congress in 2009. As a consequence, federal government saving in the recession fell faster—that is, the deficit grew faster—than our historical norm would predict.

This more-expansionary-than-usual federal fiscal policy continued through the recession and into the early part of the recovery. But in mid-2010, fiscal policy sharply reversed course. Since then, federal fiscal policy has been much more contractionary than normal. Spending has fallen sharply since 2011, and tax revenue has grown faster than usual given the weak recovery. However, the larger-than-usual deficit growth early in the recovery has offset the larger-than-usual drop in the deficit since mid-2010. As a result, overall for the recovery, fiscal policy has been only slightly more contractionary than the historical norm.

Measuring excess fiscal drag in the recovery

Figure 1
U.S. fiscal policy: Projections vs. historical norm

Source: Bureau of Economic Analysis, CBO, and authors’ calculations.

Analysts often measure the “fiscal impetus” of policy, that is, how much policy changes contribute to real GDP growth over a given period. Positive impetus indicates expansionary policy changes and negative impetus, contractionary changes. Thus, fiscal impetus can be thought of as measuring the degree to which policy is a tailwind or headwind for economic growth.

Estimating fiscal impetus has two components. The first is the change in fiscal policy as a share of GDP. The second is the multiplier, that is, the change in GDP caused by a given change in government spending or taxes. Researchers do not agree on what multipliers are most accurate. Here, we use a multiplier of one, which is near the middle of the range of empirical estimates (see Wilson 2012). Doing so allows us to focus on the effects of changes in fiscal policy on fiscal impetus.

Figure 1 shows our calculations of fiscal impetus based on actual and historical-norm estimates of noninterest spending, tax revenue, and the primary deficit. In Figure 1, the vertical line divides our results into two periods: the recovery from mid-2009 to the end of 2012, and the three years through the end of 2015. We refer to the difference between historical-norm and actual fiscal impetus as the excess drag of fiscal policy. The excess drag tells exactly how much fiscal policy is slowing the current recovery beyond the historical norm.

Panel C of Figure 1 shows the actual and the historical-norm primary deficits. The fiscal impetus based on both the actual and historical-norm deficits since the start of the recovery has been identical, −0.2 percentage point per year. In other words, federal fiscal policy has been a modest headwind to economic growth so far in the recovery, but no more so than usual given the weak pace of growth.

Fiscal drag in coming years

To assess whether fiscal policies might cause excess drag in the future, we look at projections through 2015 from the CBO for the output gap, as well as for federal spending, revenue, and the deficit. We base our calculations on the CBO’s February 2013 outlook report, which contained scenarios both with and without the sequestration budget cuts, rather than the most recent May report, which omitted the scenario without sequestration. The results for the scenario including sequestration using the May projections are very similar to those based on the February projections.

While our estimates show that fiscal policy has held back the recovery slightly to date, the effect over the next three years looks much bigger. The CBO projects that the federal deficit as a share of GDP will drop 1.4 percentage points per year over the next three years. This projection would ease slightly to 1.2 percentage points per year if sequestration spending cuts were reversed. By contrast, our calculation of the historical-norm deficit decline through 2015 is 0.4 percentage point per year based on the CBO’s output gap projections. This implies that the excess drag from the rapidly shrinking deficit would reduce real GDP growth annually by between 0.8 and 1.0 percentage point, depending on whether sequestration is reversed. Thus, with or without sequestration, fiscal policy is expected to be a much greater drag on economic growth over the next three years than it has been so far.

Surprisingly, despite all the attention federal spending cuts and sequestration have received, our calculations suggest they are not the main contributors to this projected drag. The excess fiscal drag on the horizon comes almost entirely from rising taxes. Specifically, we calculate that nine-tenths of that projected 1 percentage point excess fiscal drag comes from tax revenue rising faster than normal as a share of the economy. As Panel B shows, at the end of 2012, taxes as a share of GDP were below both their historical norm in relation to the business cycle and their long-run average of about 18%. However, over the next three years, they are projected to rise much faster than our estimate of the usual cyclical pattern would indicate. The CBO points to several factors underlying this “super-cyclical” rise, including higher income tax rates for high-income households, the recent expiration of temporary Social Security payroll tax cuts, and new taxes associated with the Obama Administration’s health-care legislation.

Conclusion

Federal fiscal policy has been a modest headwind to economic growth so far during the recovery. This is typical for recovery periods and in line with the historical relationship between the business cycle and fiscal policy. However, CBO projections and our estimate based on the countercyclical history of fiscal policy suggest that federal budget trends will weigh on growth much more severely over the next three years. The federal deficit is projected to decline faster than normal over the next three years, largely because tax revenue is projected to rise faster than usual. Given reasonable assumptions regarding the economic multiplier on government spending and taxes, the rapid decline in the federal deficit implies a drag on real GDP growth about 1 percentage point per year larger than the normal drag from fiscal policy during recoveries.

Brian Lucking is a research associate in the Economic Research Department of the Federal Reserve Bank of San Francisco.

Daniel Wilson is a senior economist in the Economic Research Department of the Federal Reserve Bank
of San Francisco.

References

Congressional Budget Office. 2013. “The Effects of Automatic Stabilizers on the Federal Budget as of 2013.” Report, March.

Lucking, Brian and Daniel Wilson. 2012. “U.S. Fiscal Policy: Headwind or Tailwind?” FRBSF Economic Letter 2012-20 (July 2).

Wilson, Daniel. 2012. “Government Spending: An Economic Boost?” FRBSF Economic Letter 2012-04 (February 6).

Yellen, Janet. 2013. “A Painfully Slow Recovery for America’s Workers: Causes, Implications, and the Federal Reserve’s Response.” Remarks at the “A Trans-Atlantic Agenda for Shared Prosperity,” conference sponsored by the AFL-CIO, Friedrich Ebert Stiftung, and IMK Macroeconomic Policy Institute, Washington, DC (February 11).

Today I’m going to talk about how the economy is doing and where I see it heading. I’ll highlight areas of strength, such as housing, and also some areas that have been holding us back, such as fiscal policy. I’ll offer my forecasts for economic growth, the job market, and inflation. And I’ll explain what the Federal Reserve is doing to keep the recovery on track and move toward the goals Congress has set for us, namely maximum employment and price stability.

Since there are a lot of CFOs here, let me go straight to the bottom line: The economy has been improving for nearly four years now. I expect this improvement to continue and to gradually gain momentum. This outlook reflects a mixture of healthy growth by households and businesses, and the dampening effects of public-sector restraint. Overall, if we were in a car, you might say we’re motoring along, but well under the speed limit. The fact that we’re cruising at a moderate speed instead of still stuck in the ditch is due in part to the Federal Reserve’s unprecedented efforts to keep interest rates low. We may not be getting there as fast as we’d like, but we’re definitely moving in the right direction.

There is indeed little doubt that the economy is on the mend. The clearest evidence can be found in housing, by far the sector hit hardest during the recession. Mortgage rates have fallen to levels rarely, if ever, seen before. Typical fixed-rate mortgages are around 3.5%, putting them in reach of millions of households. Affordable mortgages fuel demand for homes, and that pushes up sales and prices. Year-over-year, house prices are rising at around a double-digit rate.

The recovery in home prices has all sorts of beneficial effects. Increasing numbers of underwater homeowners are finding themselves on dry land again. Their properties are now worth more than their mortgages, making them less likely to default. Meanwhile, other homeowners find their mortgages have dropped below the critical 80%-of-home-value barrier. That makes it easier to refinance at today’s low rates, freeing money to spend on other things.

Homebuilders are responding to rising sales and prices. New housing construction starts have risen by more than 45% over the past year. As one of my business contacts put it, anything related to housing is doing well—furniture, appliances, even the pickup trucks workers drive to construction sites. That’s quite a turnaround from the past few years.

If you own a home, you’re probably feeling a little better about things, even if you don’t refinance. The rising value of your property—and perhaps your 401(k) as well—may be making you feel wealthier. Many people have responded to their improved finances by spending more on a range of goods and services. When we put it together, housing-related demand, improved access to credit, and the effects of increased wealth all have spurred consumer spending. The latest data show that, on a per-person basis, consumer spending has fully rebounded from its steep decline during the recession.

Meanwhile, businesses are hiring and increasing production. Manufacturing of durable goods such as motor vehicles and appliances has regained its previous level after plunging more than 25% during the recession. The recovery in durable goods also reflects the effects of lower interest rates. Of course, many businesses are still cautious. But investment in plant and equipment continues to grow moderately and is on track for further gains.

In sum, conditions in the private-sector part of the economy have improved quite a bit and continue to get better. So why then is the economy as a whole poking along instead of speeding ahead? To understand that, we need to look at what’s happening in the public sector.

Economic effects of U.S. fiscal policy

When the recession hit, state and local government tax revenue tumbled. Those governments responded by cutting spending and employment, which have still not recovered. Here’s one way to see the effect those cutbacks have had on the U.S. economy: On a per-person basis, production of goods and services is about 1% below its pre-recession peak. But if state and local government purchases today were what they were then, output per person would be roughly where it was before the recession.

The federal government is a different story. In response to the recession, the federal government cut taxes and boosted spending substantially, and that was a big plus for the economy when it was reeling. The $800 billion tax and spending package passed in 2009 was extraordinarily expansionary, and fiscal policy overall was more stimulatory than at any time since the Great Depression (see Lucking and Wilson 2012). But that stimulus has been phasing out in recent years, leading to a substantial swing in the effects of federal fiscal policy. More recently, higher taxes have come into the picture. Income tax rates were raised on upper-income Americans, and the Social Security payroll tax cut was allowed to expire. Now, on top of that, we have sequestration. In the period ahead, as the federal deficit shrinks, fiscal policy is likely to weigh on the economy even more. The federal deficit will decline by a substantial amount, on average approximately 1½% of gross domestic product every year for the three years ending in 2015, according to the Congressional Budget Office.

The United States is not the only country going through a period of budgetary restraint. In Europe, government spending has grown much slower than in other recent recoveries (see Kose, Loungani, and Terrones). Among countries that use the euro, recessions have caused tax revenue to tank, which widens budget deficits. In the face of those enormous deficits, governments have slammed the brakes on spending. But several of these economies are still contracting. Many analysts—and some policymakers as well—have begun to wonder if policies aimed at cutting deficits should be balanced with policies aimed at spurring growth.

For the United States, Europe’s struggles are a real crimp on growth. That’s because Europe is one of our major trading partners. Europeans are buying fewer American products than they would otherwise, which damps demand at a time we need it.

However, not all the news from overseas is bad. In Japan, the central bank has adopted a much more aggressive policy to promote growth. Specifically, the Bank of Japan has officially raised the level of inflation it is targeting from 1 to 2%, with a view to getting the country out of a prolonged deflationary spiral. That move may not seem like much. But in central banking circles, it’s stop-the-presses news. If the Bank of Japan succeeds in jump-starting growth, that should offset some of the economic drag coming from Europe.

Outlook for the U.S. economy

I’ve gone over some of the pluses and minuses influencing the U.S. economy so you can get a sense of what’s driving our moderate growth. What does this mean for the job market? Well, things have been getting better there as well. Since the low point for employment following the recession, the economy has added over 6 million jobs. During the past six months alone, we’ve added 1¼ million jobs. The unemployment rate of 7.5% is down 2½ percentage points from its recession peak, with nearly half a percentage point of that decline occurring in the past six months.

One reason the jobless rate has been dropping so much is that a large number of people are leaving the labor force. Many of them are reaching retirement age or going back to school. But part of this exodus appears to be people who are giving up looking for work. It’s hard to say how long these discouraged workers will stay out of the labor force. I expect many of them will return as jobs become more plentiful.

Under these circumstances, I expect the unemployment rate to decline gradually over the next few years. My forecast is that it will be just below 7½% at the end of this year, and a shade below 7% at the end of 2014. I don’t see it falling below 6½% until mid-2015. This forecast of a gradual downward trend in the unemployment rate reflects the combined effects of expected solid job gains and a return of discouraged workers to the labor force.

I also see a gradual pickup in overall economic growth. Growth of gross domestic product, which is the nation’s total output of goods and services, is likely to be relatively sluggish in the second and third quarters as sequestration begins to bite. But I expect the economy to gain momentum after that. I project that inflation-adjusted GDP will grow almost 2½% this year and 3¼% next year.

For its part, inflation is quite low, with the Fed’s preferred measure of prices rising only 1% over the past year. Wages are increasing slowly and the labor market still has considerable slack, which should restrain future wage increases. In addition, increases in the prices of imported goods and services have been subdued. And the public continues to expect low inflation. I expect that the decline in inflation will prove to be temporary, and that inflation will climb slowly, but stay below the Fed’s 2% longer-run target over the next few years.

Federal Reserve policies

Earlier I noted that the economy’s improved performance stemmed in part from Federal Reserve policies. Let me describe what we’ve done and then shift to what we might do in the future. During the recession, the Fed acted quickly and aggressively. We pushed our benchmark short-term interest rate close to zero at the end of 2008. But that move was not nearly enough to offset the damage caused by the financial crisis and the housing collapse. We couldn’t push short-term interest rates any lower. So we had to devise new, unconventional ways to stimulate the economy.

Broadly, our unconventional policies have fallen into two categories: The first involves what we say, the second, what we do. As far as what we say is concerned, our approach is centered on what is known as forward guidance. Under forward guidance, the Fed’s policy committee releases public statements about the likely stance of policy in the future. The aim is to reduce public confusion and uncertainty about future Fed policy, and thereby help us achieve our policy goals. For example, let’s take the federal funds rate, our benchmark short-term interest rate. If the Fed’s policy committee states that it expects the federal funds rate to remain exceptionally low for an extended period, that will also drive down longer-term interest rates right away. And those longer-term rates have a lot to do with whether a young family buys a house or a car, or a business builds a new factory.

The Federal Reserve’s policy committee, the Federal Open Market Committee or FOMC, has issued forward guidance that spells out specific economic conditions that serve as thresholds for considering increases in the federal funds rate. In particular, the Committee’s policy statements have specified that we expect to keep the federal funds rate exceptionally low at least as long as, one, “the unemployment rate remains above 6½%”; two, “inflation between one and two years ahead is projected to be no more than a half percentage point above the Committee’s 2% longer-run goal”; and three, longer-term inflation expectations remain in check (see Board of Governors 2012).

With this forward guidance in place, members of the public can adjust their expectations for future Fed policy as new information on the economy becomes available. They don’t need to wait for the Fed to issue a new statement. For example, a slowdown in economic growth might cause the public to think that the prospect of reaching a 6½% unemployment rate was falling further back in time. They would then expect the Fed to wait longer to raise the federal funds rate, which would prompt them to push long-term interest rates down. And those lower long-term rates would help us achieve our monetary policy goals.

The second major unconventional policy category involves what we do. Here I am speaking of our program to buy a total of $85 billion of agency mortgage-backed securities and longer-term Treasury securities each month, often referred to in the financial press as QE3. Under the current and earlier asset purchase programs, we’ve bought more than $3 trillion in longer-term Treasury and mortgage-related securities. Fed purchases boost demand for these securities, bidding up their prices and lowering their yields. Lower Treasury and mortgage yields spill over into other markets, lowering longer-term rates across the board (see Williams 2011, 2012).

The lower rates that stem in part from forward guidance and quantitative easing have big benefits for the economy. Take a homeowner with a $250,000 mortgage. With the decline in longer-term interest rates since 2009, that homeowner might be paying around $3,000 less on a mortgage each year, if he or she is able to refinance.

Similar to our forward guidance on the federal funds rate, the FOMC has linked our asset purchases to the outlook for the economy. Specifically, we’ve said we expect to continue buying longer-term Treasury and mortgage securities until the outlook for the job market improves substantially, provided inflation remains contained. In the statement issued by the FOMC May 1, we said we would adjust our securities purchases to ensure that monetary stimulus is at a level appropriate for economic conditions. That has been a principle of our ongoing open-ended securities purchase program since we started it in September. It means we will alter the size, pace, and composition of our purchases as necessary as the economy evolves.

Figure 1
Nonfarm payroll employment

Source: Bureau of Labor Statistics.
Note: Seasonally adjusted.

Figure 2
Unemployment rate

Sources: Bureau of Labor Statistics and FRBSF staff.
Note: Seasonally adjusted.

So, do economic conditions suggest we need to change the $85 billion in monthly securities purchases we’re currently making? To answer that, it’s useful to look at what conditions were when we launched the program last September. At that time, the economy was flashing warning signals. As Figure 1 shows, the pace of improvement in the labor market had slowed, with payroll growth declining. Moreover, the unemployment rate appeared to be stalled at about 8%, and we at the San Francisco Fed did not see the rate falling below that level till the final quarter of 2013, as shown by the blue dotted line in Figure 2.

Since then, the labor market has improved considerably. The pace of job growth has picked up, averaging over 200,000 jobs per month over the past six months. And the unemployment rate has come down. The red line in Figure 2 shows that, in the seven months since our latest securities purchase program began, the unemployment rate has fallen faster than we had expected.

It’s clear that the labor market has improved since September. But have we yet seen convincing evidence of substantial improvement in the outlook for the labor market, our standard for discontinuing our securities purchases? In considering this question, I look not only at the unemployment rate, but also a wide range of economic indicators that signal the direction the labor market is likely to take.

Economists at the San Francisco Fed have identified a very good set of indicators for this purpose that tell us what labor market conditions are likely to be six months in the future. They include well-known labor market measures, such as private payroll employment growth and initial unemployment insurance claims. But they also include measures that are not as well known, such as growth in temporary help employment and survey data on the share of households that find jobs hard to get.

Consistent with the payroll and unemployment data I mentioned earlier, most of these indicators look healthier than they did in September. What’s more, nearly all of them are signaling that the labor market will continue to improve over the next six months. This is good news. But it will take further gains to convince me that the “substantial improvement” test for ending our asset purchases has been met. However, assuming my economic forecast holds true and various labor market indicators continue to register appreciable improvement in coming months, we could reduce somewhat the pace of our securities purchases, perhaps as early as this summer. Then, if all goes as hoped, we could end the purchase program sometime late this year. Of course, my forecast could be wrong, and we will adjust our purchases as appropriate depending on how the economy performs.

It’s important to stress that even if we slow the pace of our purchases, it does not mean we would be tightening monetary policy or stepping back from our commitment to provide strong monetary support as the economy recovers. The stance of monetary policy would still be extremely stimulatory. In terms of our asset purchases, the evidence shows that the stimulus we are providing depends on the size of our balance sheet, not the rate at which we’re buying assets. So even when we reduce or halt new purchases, we’ll still have trillions of dollars of longer-term securities on our balance sheet exerting downward pressure on interest rates.

Of course, eventually we will adjust our policy stance back toward normal levels. When we do, it will be because changing circumstances have made an adjustment the best way to lead us toward our mandated goals of maximum employment and price stability. We recognize that much is uncertain when it comes to the economy, and we’ve thought a great deal about how best to manage our exit from these unconventional policies. I am confident that we will succeed in doing so.

John C. Williams is president and chief executive officer of the Federal Reserve Bank of San Francisco.

References

Board of Governors of the Federal Reserve System. 2012. “Press Release.” December 12.

Kose, M. Ayhan, Prakash Loungani, and Marco E. Terrones. 2013. “The Great Divergence of Policies.” Box 1.1 in World Economic Outlook: Hopes, Realities, and Risks. Washington, DC: International Monetary Fund, pp. 32–35.

Lucking, Brian, and Daniel Wilson. 2012. “U.S. Fiscal Policy: Headwind or Tailwind?” FRBSF Economic Letter 2012-20 (July 2).

Williams, John C. 2011. “Unconventional Monetary Policy: Lessons from the Past Three Years.” FRBSF Economic Letter 2011-31 (October 3).

Williams, John C. 2012. “The Federal Reserve’s Unconventional Policies.” FRBSF Economic Letter 2012-34 (November 13).

Assessing the contribution of cyclical factors and the likelihood of a reversal or slower decline in labor force participation is difficult based on aggregate labor market data alone. Such data cannot perfectly distinguish between long-term trends and shorter-term cyclical factors, particularly given the severity of the labor market dislocation during the past recession. To assess the role of cyclical factors in the current recovery, we examine state-level variation in the relationship between changes in the labor force participation rate and changes in employment over several business cycles.

Aggregate labor force participation rate trends

The labor force participation rate is defined as the percentage of the civilian noninstitutional population 16 and over working or looking for work. It is largely determined by demography, most notably the share of the adult population of prime working age, typically 25 to 54. Younger people often are in school and older people often are retired, reducing their respective participation rates. The rate is also determined by long-term socioeconomic trends, such as wider entry of women into the labor force starting in the 1960s; changes in income and wealth; and the availability and generosity of government benefit programs (Daly and Regev 2007; Daly, Hobijn, and Kwok 2009).

In addition, the labor force participation rate may reflect short-term cyclical influences. Research suggests a weakly pro-cyclical relationship between the aggregate participation rate and broad economic conditions. Participation tends to rise a bit during expansions when jobs are plentiful and edge down in recessions when jobs are scarce (Van Zandweghe 2012).

Figure 1
Labor force participation rate

Sources: BLS/Haver Analytics.
Note: Gray bars show NBER recession dates.

Figure 1 shows the aggregate U.S. labor force participation rate since 1948. An upward trend during most of the post–World War II period appears to have reversed around 2000 when a downward trend emerged. That trend intensified during the 2007–09 recession. Researchers have identified a number of factors that may account for the shift. They include the baby boom cohort moving past their prime working-age years; the stabilization of women’s labor force participation rates; more younger working-age people enrolling in school; and increased use of some social benefit programs, notably disability insurance (Daly and Regev 2007; Daly et al. 2009; Aaronson, Davis, and Hu 2012).

Because the downward trend in participation started around 2000, it is difficult to identify the portion of the decline since 2007 that is cyclical and likely to be reversed as the labor market recovery continues. Researchers have used several approaches to tease out the cyclical component. These include comparisons across demographic groups; across different categories of unemployed workers and people out of the labor force who want work; and of actual outcomes versus hypothetical outcomes based on adjusting demographic information (Aaronson et al. 2012; Daly et al. 2012; Van Zandweghe 2012; Hotchkiss and Rios-Avila 2013). These studies have found a potentially large cyclical component in the recent participation decline.

State-level evidence

We assess cyclical fluctuations in the national labor force participation rate by examining differences in cyclical labor market conditions and labor force participation rates across states (see also Erceg and Levin 2013). Any correlations between changes in labor market conditions and participation rates at the state level are likely to be mirrored at the national level as well. This assumption is not directly testable, but the approach is potentially useful as an alternative to other methods.

We use state-level payroll employment growth to measure cross-state differences in labor market conditions. Payroll growth is preferable to state unemployment rates because it is measured separately from the state’s labor force participation rate. Since the unemployment rate varies with the participation rate, cross-state analysis of the relationship between unemployment and participation could be contaminated. The payroll data are from the U.S. Bureau of Labor Statistics (BLS) monthly survey of employers, while participation rates are from the BLS monthly household survey.

Our analysis is based on state-level changes in payroll employment and labor force participation during the downturn and recovery periods of the recessions of 1981–82, 1990–91, 2001, and 2007–09. We define a downturn as the period between the peak and trough in national payroll employment. These periods are similar to official recession dates identified by the National Bureau of Economic Research (NBER), but they can differ. For example, we identify the most recent downturn as occurring from January 2008 to February 2010, while NBER dates the overall recession from December 2007 to June 2009. The recovery is defined as dating from the end of the national recession. The data for the 1990–91 and 2007–09 recession and recovery periods were adjusted to minimize the effects of population estimate revisions and the 2010 surge in hiring of census workers. All calculations are weighted by the relative size of each state’s labor force. This places greater weight on more populous states, capturing the greater precision of their employment and labor force estimates, and more accurately measuring the estimated relationship between payroll growth and participation. When we weight all states equally, correlations are weaker.

Figure 2
State employment, participation rates during recession

Changes from January 2008 to February 2010

Sources: BLS/Haver Analytics and authors’ calculations.
Note: Line weighted by state’s fraction of total labor force over period.

Table 1
Correlation between changes in employment and participation rates

Figure 2 displays the relationship between the percentage changes in state-level labor force participation rates and payroll employment in the most recent downturn. The figure shows wide cross-state variation in the extent of job loss and changes in participation. The upward-sloping blue line shows that changes in employment and participation are positively related across states. Larger declines in employment are associated with larger declines in labor force participation rates. This systematic relationship at the state level between the severity of employment losses and the decline in participation suggests that the drop in the national participation rate may also have an important cyclical component. Our results reinforce the findings of other researchers who have found evidence of cyclicality in the labor force participation rate.

Table 1 shows the results of an analysis of the correlation between changes in payroll employment and labor force participation in the past four recessions and recoveries. The degrees of correlation measured by this statistical analysis broadly confirm the results illustrated by the upward sloping line in Figure 2. The positive relationship in the most recent downturn shown in Figure 2 is generally, but not invariably, evident in past downturns and recoveries. The main exception is the 2001 recession. Our analysis finds little or no systematic cross-state relationship between changes in employment and participation in that episode. The 2001 recession may have been different from the other recessions in that it was brief and mild, and its impact was concentrated in a few sectors and states.

Although the 2007–09 downturn exhibits a strong positive relationship between state-level changes in employment and participation, the recovery so far does not. This calls into question our interpretation that much of the recent participation decline is cyclical and likely to reverse. However, the current weak correlation between changes in employment and labor force participation could reflect employment’s relatively modest recovery to date. The economy has been expanding for a sustained period. But, as of March 2013, we have recovered only 67% of total jobs lost during the downturn. Thirty-seven months after the employment trough in past recoveries, employment greatly exceeded the pre-recession peak.

To put the current and past recoveries on more equal footing, we calculated correlations between changes in payroll employment and participation rates for the past four recoveries over the periods it took for 67% of jobs to be regained. The last column in Table 1, labeled partial recovery, shows the results. For the 1981–82 and 1990–91 recessions, the partial recovery correlations are much smaller than those for the full recovery and are not statistically significant. Thus, we may not be deep enough into the current recovery for the typical positive relationship between participation and employment growth to emerge.

In the recoveries from the 1981–82 and 1990–91 recessions, the positive relationship did not emerge until the economy had passed the previous employment peak by a substantial margin. These results are not definitive, but they reinforce other research that finds labor force participation at the state and national levels may bounce back or decline less rapidly as the current recovery gains strength. If exceeding the pre-recession employment peak is a prerequisite for the correlation to become significant, it may take several years before the relationship is evident.

Conclusion

The U.S. labor force participation rate has declined sharply since 2007, intensifying a downward trend that has been evident since about 2000. Distinguishing between long-term influences on the participation rate, such as demographics, and short-term cyclical effects is important because it helps us understand and predict the future path of macroeconomic variables such as the unemployment rate. Using state-level evidence on the relationship between changes in employment and labor force participation across recessions and recoveries, we find evidence, reinforcing other research, that the recent decline in participation likely has a substantial cyclical component. States that saw larger declines in employment generally saw larger declines in participation. A similar positive relationship was evident in past recessions and recoveries. In the current recovery, it will probably take a few years before cyclical components put significant upward pressure on the participation rate because payroll employment is still well below its pre-recession peak.

Leila Bengali is a research associate in the Economic Research Department of the Federal Reserve Bank of San Francisco.

Mary Daly is a group vice president and associate director of research in the Economic Research Department of the Federal Reserve Bank of San Francisco.

Rob Valletta is a research advisor in the Economic Research Department of the Federal Reserve Bank of San Francisco.

References

Aaronson, Daniel, Jonathan Davis, and Luojia Hu. 2012. “Explaining the Decline in the U.S. Labor Force Participation Rate.” FRB Chicago, Chicago Fed Letter 296 (March).

Daly, Mary, Early Elias, Bart Hobijn, and Òscar Jordà. 2012. “Will the Jobless Rate Drop Take a Break?” FRBSF Economic Letter 2012-37 (December 17).

Daly, Mary, and Tali Regev. 2007. “Labor Force Participation and the Prospects for U.S. Growth.” FRBSF Economic Letter 2007-33 (November 2).

Daly, Mary, Bart Hobijn, and Joyce Kwok. 2009. “Labor Supply Responses to Changes in Wealth and Credit.” FRBSF Economic Letter 2009-05 (January 30).

Erceg, Christopher J., and Andrew T. Levin. 2013. “Labor Force Participation and Monetary Policy in the Wake of the Great Recession.” Manuscript, International Monetary Fund, April 9.

Hotchkiss, Julie, and Fernando Rios-Avila. 2013. “Identifying Factors behind the Decline in the U.S. Labor Force Participation Rate.” Business and Economic Research 3(1, March), Macrothink Institute, pp. 257–275.

Van Zandweghe, Willem. 2012. “Interpreting the Recent Decline in Labor Force Participation.” FRB Kansas City Economic Review, first quarter.

The more things change, the more they stay the same

Can you identify when the following events took place? Early in the year, the stock of a nonbank financial institution widely used as collateral in certain transactions drops by more than half. At first, financial markets shrug it off. But, during the summer, with market distress apparently in check, two new financial challenges arise. Shortly after that, a nonbank financial institution fails and panic strikes financial markets. Meanwhile, a giant financial company is also in trouble, but considered salvageable. A deal is struck. Market participants bet that the rescue will keep other financial dominoes from falling. Disaster is avoided, but not without major damage to the real economy in the form of a protracted recession.

If you thought this scenario describes the 2008 financial crisis, you would be right. You would have identified Bear Stearns as the company whose stock took a hit early in the year, Fannie Mae and Freddie Mac as the two challenges that surfaced in the summer, Lehman Brothers as the nonbank financial institution that collapsed shortly after that, and AIG as the financial giant that was rescued. However, the scenario also describes events that took place a century earlier. Replace Bear Stearns with Union Pacific, Fannie Mae and Freddie Mac’s failures with the crash of copper prices and New York City bonds, Lehman Brothers with the Knickerbocker Trust Company, and AIG with the Trust Company of America, and you would precisely identify the panic of 1907.

The events surrounding both crises were remarkably similar. But the institutional responses to these events were vastly different. The lack of a central bank in 1907 meant there was no public lender of last resort. Instead, J. Pierpont Morgan, a private investor and founder of J.P. Morgan & Co., orchestrated the rescue effort in 1907. In 2008, it was federal government agencies, especially the Federal Reserve and the Treasury Department, that responded to the crisis.

Origins of the Federal Reserve

Recessions originating from a financial event were common in the late 19th and early 20th centuries. Many stemmed from banking panics. Figure 1 provides a global historical perspective. We calculate by decade the number of countries that experienced financial crises among a sample of 17 industrialized economies representing more than half of global GDP during the past 140 years (for details, see Jordà, Schularick, and Taylor 2012).

Figure 1
Countries experiencing financial crises by decade

Source: Jordà et al. (2012).

Figure 1 shows a notable downward global trend in the incidence of these highly disruptive events, with the conspicuous exceptions of the Great Depression and the Great Recession of 2007–09. In the United States, the rate of banking crises declined markedly after the 1913 creation of the Federal Reserve System. Other than the Great Depression and Great Recession, the only significant banking crisis of the past century was the savings and loan crisis. By contrast, ten significant banking crises occurred in the 19th century.

The panic of 1907 and the resulting recession are generally credited with providing the catalyst for the creation of the Federal Reserve System. When the Federal Reserve was chartered, the United States had been without a central bank for about 70 years. Congress chartered The First Bank of the United States in 1791 during the Washington presidency, under the guiding hand of Secretary of Treasury Alexander Hamilton. However, its 20-year charter was allowed to expire in 1811. Then, under President Madison, the Second Bank of the United States was created in 1817 for another 20-year period. Once again, the charter was allowed to expire amid President Jackson’s strong opposition to the central bank.

For the next 70 years, a period known as the Free Banking Era, no national public authority policed the banking system. Banks, which earned their livelihood from the creation of money and credit, were entrusted to regulate the system’s excesses. This period was punctuated by eight banking crises (Reinhart and Rogoff 2009).

(The Federal Reserve Bank of San Francisco exhibits currency privately issued by banks during this period as well as the rare and valuable Grand Watermelon note and an 1886 $1 silver certificate with a portrait of Martha Washington, the only woman ever to appear on American currency, see http://www.frbsf.org/currency).

Lender of last resort and interest rate policy

Charles Kindleberger’s 1978 classic Manias, Panics and Crashes asked whether a lender of last resort moderates the business cycle (see Kindleberger, Aliber, and Solow, 2011). His tentative answer was “that a lender of last resort does shorten the business depression that follows a financial crisis.” Taking Kindleberger’s proposition more broadly, we compare a U.S. recession without a central bank with a downturn in which the Federal Reserve was actively working to protect the financial system and stimulate the economy: the 1907 Panic and the 2007–09 financial crisis and recession. Figure 2 compares real GDP per capita in the United States around those two events.

Figure 2
Real GDP per capita cumulative change

Source: Jordà et al. (2012).

Real GDP per capita followed very similar paths in each crisis. Declines in the first year of recession intensified rapidly in the second. But, at the trough, real GDP per capita had dropped by more than 10% in the recession associated with the 1907 Panic compared with about 5% in the Great Recession. Recovery from the trough was somewhat quicker for the 1907 Panic. But, at the five-year mark, the cumulative change in real GDP per capita was very similar.

We can also compare the path of unemployment after each crisis. If unemployment had increased at the same rate in the second year of the Great Recession as during the downturn associated with the 1907 Panic, a rough estimate indicates that the total number of unemployed people in 2009 would have been about 2.5 million higher than the 14.3 million actually observed. It is natural to wonder whether the recession associated with the 1907 Panic would have been shallower had we been able to transport the Federal Reserve back in time.

Although J.P. Morgan acted as a de facto lender of last resort, the absence of a de jure government institution that could play this role probably aggravated the panic. However, counterfactual historical analysis is difficult because we cannot conduct a controlled experiment. Many factors contributed to the actual course of events. The absence of a central bank was far from the only distinction between the recession associated with the panic of 1907 and the Great Recession of 2007–09.

Modern statistical methods provide one way to tease out the effects of different macroeconomic factors. Using a comprehensive historical database, Jordà et al. (2012) examine the determinants of financial crises and their cyclical patterns. Their model explores the dynamics of the economy once a financial crisis strikes, permitting comparison of events across eras and countries.

The experiment we conduct focuses on monetary policy. In 1907, no official monetary authority existed and short-term interest rates were set by the market. By contrast, during the recent recession, the Federal Reserve aggressively lowered its benchmark interest rate from 5.25% to near zero by December 2008. Would the drop in real GDP per capita associated with the 1907 Panic have been less severe if interest rates had been cut the same way?

Using the Jordà et al. (2012) model, we can compare the actual path of short-term interest rates on government securities with the model’s prediction for those rates. Although these rates may not give an accurate picture of the rates actually levied on businesses and households, they give a good sense of changes in overall credit conditions. Comparing the model’s predictions with actual outcomes provides an indirect measure of the effects of policy intervention.

After the first year of recession, short-term interest rates were about one percentage point above what the model predicts for the 1907 episode, but about two percentage points below the model’s prediction for the Great Recession. This appears to reflect the Fed’s aggressive rate-cutting response to the financial crisis. In fact, this general pattern of rates higher than predicted for the 1907 crisis and lower than predicted for the Great Recession characterizes the five-year windows for both periods.

Using the difference between interest rates predicted by the model and actual rates, we can impose the effect of the Fed’s rate reductions in 2007 and 2008 on the 1907 crisis. To do this, suppose that the interest rate prediction errors in 1907 were the same direction and size as those in the Great Recession. How would the path of real GDP per capita after the 1907 Panic have changed? If interest rates had been cut in the same way as they were in 2007 and 2008, the contraction in real GDP per capita would have been about two percentage points less in the second year. This may seem small, but two percentage points is about the average loss of per capita real GDP during a typical recession. If per capita GDP had fallen by an additional two percentage points in 2009, over one million more jobs would have been lost. The stimulus package passed in 2009 did not take full effect until 2010. So differences in the recovery path cannot be explained by differences in fiscal policy.

Conclusion

The duties and authority of the Federal Reserve System, established by the Federal Reserve Act of 1913, have evolved continuously to keep pace with an ever-more-complex financial system. Financial crises have become less frequent since 1913, but they have not been completely eliminated. The presence of the Fed as monetary authority and lender of last resort distinguishes the financial crisis of 2007–08 from the 1907 Panic. In the recession associated with the recent financial crisis, losses in economic activity and employment were less severe. When other functions not examined here are considered, such as price stability, the success of the Federal Reserve in mitigating recessions is clearer. The financial landscape of the next 100 years will continue to evolve. We can be certain that, in its second century, the Fed will have to meet new challenges yet unknown.

Early Elias is a research associate in the Economic Research Department of the Federal Reserve Bank of San Francisco.

Òscar Jordà is a research advisor in the Economic Research Department of the Federal Reserve Bank of San Francisco..

References

Jordà, Òscar, Moritz Schularick, and Alan M. Taylor. 2012. “When Credit Bites Back: Leverage, Business Cycles and Crises.” FRB San Francisco Working Paper 2011-27.

Kindleberger, Charles P., Robert Z. Aliber, and Robert Solow. 2011. Manias, Panics and Crashes: A History of Financial Crises. Hoboken, NJ: John Wiley & Sons, 6th edition.

Reinhart, Carmen M., and Kenneth Rogoff. 2011. This Time Is Different: Eight Centuries of Financial Folly. Princeton, NJ: Princeton University Press

Figure 1
Commercial real estate investment over business cycles

Note: Shares of real GDP indexed to 1 at cyclical peak.

The commercial property downturn in part reflects how the slump in the broader economy led to a deterioration of real estate fundamentals, such as rental price appreciation and vacancy rates. The magnitude of the collapse in new construction was probably also due to the extraordinary developments on the pricing and funding side of the commercial real estate sector. Commercial property prices fell about 40% from late 2007 to early 2010. This shock to real estate collateral values led to a sharp contraction in funding for commercial real estate projects. Commercial real estate loans outstanding fell 18%, and securitization of new commercial mortgages seized up.

Figure 1 could be read as indicating that the entire commercial real estate market is still seriously depressed. However, the reality is more nuanced. First, the commercial real estate market consists of both new and existing properties. It’s true that builders are not adding much new space. But there are signs of a rebound in the market for existing properties. Second, drilling down below the aggregate statistics, commercial real estate is performing differently both within and across geographical markets. Furthermore, owners of properties that are completed and fully leased have access to credit on very favorable terms. By contrast, conditions are different for more marginal properties that are not leased up or producing reliable cash flows.

Figure 2
CMBS spreads

Let’s examine the first point, that conditions in the existing commercial property market are better than might be predicted based on the level of new nonresidential construction. One piece of evidence comes from the risk premiums that investors in commercial mortgage-backed securities (CMBS) require, which are reflected in the interest rate spreads over comparable risk-free rates. Figure 2 plots the path of the spreads of an index of AAA-rated CMBS yields over 10-year Treasury securities. Spreads on the senior CMBS tranche, which are the safest claims, are shown by the solid blue line. These spreads spiked in 2008 during the financial crisis, but have since moved back down to levels in effect before the crisis. All the same, concerns about risk are still evident in the CMBS market. The spreads on the riskier junior tranche of the AAA-rated CMBS index, indicated by the dashed red line, have not recovered as much as for senior bonds. Moreover, these spreads shot up again, along with all other risk spreads, in response to the European sovereign debt crisis.

Commercial real estate investments typically require a high proportion of borrowed funds. Access to and terms for credit figure importantly in how able and willing investors are to pay for properties. The easing of pricing for commercial real estate debt has helped fuel a mild lending recovery. Securitization of commercial real estate loans is nowhere near its level before the recession, but the pace of issuance has begun to revive. Likewise, commercial bank lenders have returned to the market, and the stock of bank nonresidential real estate loans has ticked up.

Valuation measures in commercial real estate

One common metric for valuing commercial real estate is the capitalization rate, or cap rate. It is defined as the ratio of the expected annual net operating income on a property to the price of the property. The concept is similar to the earnings yield on a stock. Net operating income changes slowly, so much of the variation in cap rates over time is due to changing property valuations.

As should be expected, interest rates, cap rates, and commercial real estate valuations move closely together. A basic principle of finance is that prices are the present value of future expected cash flows. Those prices depend critically on what discount rate is applied to these cash flows. As interest rates fall, the rate at which the cash flows on commercial properties are discounted also falls, pushing commercial real estate prices up.

Hobijn, Krainer, and Lang (2011) investigated the behavior of cap rates in different regional markets and different property categories, including offices, retail, industrial, and multifamily residential. Their goal was to explain what drives cap rates, that is, to what extent cap rates reflect discount rates and expected future cash flows respectively. They constructed a weighted index of cap rates from metropolitan markets across the country using a statistical technique called principal components analysis. They found that this weighted cap rate index moved closely with the level of interest rates. This suggests that changes in interest rates, which occur nationwide, lead to changes in commercial real estate discount rates across all local markets.

By contrast, after accounting for the interest rate component in the statistical analysis, other measures of real estate fundamentals, such as regional unemployment rates, have weak relationships with metropolitan cap rates. This is not to say that cap rates have no relationship to any economic variable except interest rates. Cap rate levels still vary over time with idiosyncratic features of local economies or individual properties. It is simply that most of the common variation of cap rates across markets can be attributed to the movement of interest rates over time.

Figure 3
Cap rate comparisons for commercial real estate

Source: CBRE.
Note: Spread=Cap rate – 10-year Treasury inflation-protected securities yield.

A close look at commercial real estate fundamentals underscores the critical role interest rates play in determining cap rates. For most classes of commercial real estate, vacancies and rents have yet to recover significantly from the effects of the recession. But, as Figure 3 shows, for office, industrial, retail, and multifamily housing properties, cap rates, like interest rates, are at historical low points. This suggests that low interest rates are one of the only things currently supporting commercial real estate prices. The main exception is multifamily housing, which is seeing rising rents as well as historically low cap rates. Multifamily housing has undoubtedly benefited from the depressed demand for owner-occupied housing.

The improvements in cap rates have also been pervasive across different regional markets. Figure 4 shows that cap rates in primary metropolitan markets fell significantly from the first quarter of 2010 to the third quarter of 2012. This makes sense given the importance of interest rates for commercial real estate valuations. Of course, interest rates are determined in global financial markets. Borrowers with commercial property in different regional markets compete for funding in the broad financial market. Changes in interest rates should filter down to property markets everywhere. However, despite the nationwide improvement in commercial real estate, significant regional variation exists.

Figure 4
Distribution of regional office cap rates

Source: Metropolitan statistical area data from CBRE.

Figure 4 shows the geographic dispersion in cap rates. For example, average cap rates in San Francisco are currently close to 4%, while cap rates in Detroit are closer to 7%. In other words, investors value a dollar of earnings on commercial property in San Francisco at a multiple of 25. But they are only willing to pay about 14 times earnings for property in Detroit. Similarly, within markets, cap rates vary based on property classification. Cap rates on both Class A and Class B properties have generally come down over the last two years. But, even within the same metropolitan area, significant gaps in value are found between higher- and lower-quality properties. This undoubtedly reflects different local economic conditions and different expectations for future earnings growth even for properties within the same geographic market. These valuation disparities suggest that there still are very large differences in opportunities for different kinds of projects to get funding.

Conclusion

The improvement in commercial real estate cap rates appears to be largely the result of the recovery in credit markets. Cap rates are close to their historic lows for most property classes. At the same time, other commercial real estate fundamentals are still weak. This apparent disconnect—low cap rates and weak fundamentals—has prompted some observers to question the Federal Reserve’s low interest rate policy. The concern is that low rates may be boosting commercial real estate prices excessively. At this point, this concern does not appear to be warranted. It’s true that cap rates are at historic low levels. But it’s important to compare cap rates with other financial market yields rather than with cap rates during other periods. Many market interest rates are at or near historic lows, so low cap rates are not anomalies.

To elaborate, the red dashed lines in Figure 3 show the spread between cap rates and the yield on inflation-protected Treasury securities (TIPS). TIPS yields represent a real interest rate since they adjust to inflation. Thus, they are an appropriate benchmark for cap rates, which are based on cash flows that also adjust to inflation. Based on current cap rates, commercial real estate yields are very low. However, other benchmark bond market yields are even lower, including nominal yields that don’t adjust to inflation, such as the 10-year Treasury note or risky corporate bonds. This suggests that low cap rates are natural in a low interest rate environment. In itself, that does not tell us whether low interest rates are leading to excessive commercial real estate pricing. However, it does support the notion that improvements in commercial real estate are part of a broader healing process taking place throughout the economy.

John Krainer is a senior economist in the Economic Research Department of the Federal Reserve Bank of San Francisco.

References

Hobijn, Bart, John Krainer, and David Lang. 2011. “Cap Rates and Commercial Property Prices.” FRBSF Economic Letter 2011-29 (September 19).

]]> http://www.frbsf.org/economic-research/publications/economic-letter/2013/april/commercial-real-estate-low-interest-rates/feed/ 0 http://www.frbsf.org/economic-research/publications/economic-letter/2013/april/job-economic-growth-california/ http://www.frbsf.org/economic-research/publications/economic-letter/2013/april/job-economic-growth-california/#comments Mon, 15 Apr 2013 07:00:22 +0000 cpatton http://www.frbsf.org/economic-research/?post_type=frbsf_publications&p=28638 How does California’s economic performance compare with that of other states? Consider two of the main barometers of state economic performance: economic output and jobs. Typically, when a state’s economy expands, we expect the number of jobs to grow to the same extent. But from 1990 to 2011, California’s growth did not follow this pattern. Economic output in California grew faster than in many states, while job growth was slower than most states.

This Economic Letter provides an empirical description of California’s economic performance compared with other states, focusing on the metrics of output and jobs. It draws on results from a large research project, Compare50.org, which provides a rich, multidimensional database on individual state economic performance (see Neumark and Muz, 2013, and view the project at http://www.Compare50.org). The Letter also explores two possible reasons why California has performed differently than the rest of the United States. One reason is that employment in the state has shifted to high-wage industries with high levels of productivity that require fewer workers. A second possible reason, that higher wages are needed to cover higher housing costs, cannot explain the difference between economic and job growth.

Economic growth and job growth

Economic growth at the state level is commonly measured using overall state economic output, or gross state product (GSP). We look at inflation-adjusted, or real, GSP to avoid the effects of rising prices. And we consider real GSP on a per capita basis to measure the growth in economic resources per person, rather than growth that comes from a rising population.

Figure 1
Comparison of per capita real GSP growth

Sources: Bureau of Economic Analysis and authors’ calculations; see Neumark and Muz (2013).

In assessing whether a single state’s economic performance is strong or weak, it makes sense to compare performance across states because business cycles affect the whole nation. We compare California’s performance with the rest of the United States, and with other western states, including Alaska, Arizona, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming. Figure 1 shows growth in real per capita GSP for California, the western states excluding California, and the United States, excluding California but including the District of Columbia. Since the early 1990s, real GSP growth in California has followed a more distinct cyclical pattern than in the western region or the rest of the United States. Overall, California’s economic growth was only slightly lower than that of other states. Of course, economic output in all three areas fell during the Great Recession from 2007–09.

For the whole sample period 1990 to 2011, California ranked 29th in output growth, meaning that 22 states had lower growth and 28 had higher growth. If we exclude the Great Recession and consider data only through 2007, California’s growth compares more favorably with that of other states, ranking 19th. Thus, excluding the Great Recession, California’s economic growth outpaced about 60% of states in the nation. Still, each of the three recessions of the past 20 years—for different reasons—have been more severe in California than in the rest of the nation.

Our second measure of a state’s economic health is job growth. To measure it, we use data collected from employers through the Quarterly Census of Employment and Wages (QCEW) conducted by the U.S. Bureau of Labor Statistics. The QCEW data provide a reliable measure of job growth over the long term because they capture nearly all employment in the U.S. economy, not just a sample.

Figure 2
Comparison of overall job growth

Sources: QCEW and authors’ calculations; see Neumark and Muz (2013).

Figure 2 shows California employment growth relative to that in the rest of the United States and the western region. Comparing it with Figure 1, it is evident that job growth largely mirrors output growth. However, the figures also show that, unlike output growth, job growth in California lagged behind the United States and the rest of the western region during the mid-2000s boom. This implies that, to some extent, overall economic output spurted ahead in California during this period without the usual accompanying increase in jobs. A similar pattern is also apparent in the late 1990s when California’s output boom was much more pronounced than its jobs boom.

Looking at job growth over the entire 1990 to 2011 period provides more evidence for this pattern. In contrast to GSP growth, California’s job growth was relatively anemic compared with that in other states. For the whole period, only 11 states had slower job growth. This pattern holds even if we look at the data only through 2007. Only 12 states reported slower job growth. Thus, California was in the bottom quarter of states for job growth, but near the median for output growth.

Why did output growth outstrip job growth in California?

A number of factors could explain why output growth has outstripped job growth in California. Perhaps the most natural explanation is that employers in the state have hired relatively more higher-wage workers. Higher-wage workers tend to be more productive, that is, they produce more output per hour of work. As a result, California could have registered disproportionately large growth of output for the number of jobs created.

One reason California employers may have hired more high-wage workers than employers in other states is high housing costs. All else the same, to afford more expensive housing, workers must earn more. If worker earnings have to compensate for the higher cost of housing, employers may use fewer workers overall, substituting away from lower-skill, lower-wage workers.

These links can get complicated. High housing costs could conceivably reflect a more productive economy, not the other way around. A more productive economy enables employers to pay more, and a large number of highly paid employees can drive up house prices. At the same time though, as the area’s economy improves, it could offer more amenities that workers like, enabling employers to pay them less. Alternatively, such factors as the mix of industry could have led to more hiring of high-skill, high-earnings workers in California. For example, some evidence suggests that California’s relatively faster growth in high-wage jobs may have been partly fueled by growth in technology- and information-intensive industries.

Figure 3
Housing costs and changes in high-wage employment

Sources: U.S. Department of Housing and Urban Development, QCEW, and authors’ calculations; see Neumark and Muz (2013).

Figure 3 shows that California’s housing costs over the period were 40% higher than in the rest of the nation. The figure also shows that the share of employment in high-wage industries in California grew more than 0.6 percentage point faster per year than in the rest of the United States. High-wage industries are identified using QCEW data for employment in the top third of industries ranked according to average wages per employee across the years of our study.

Do these two factors—higher housing costs and faster growth in high-wage employment—help explain the gap between economic growth and job growth in California? To answer this, we check whether other states that experienced higher housing costs and faster high-wage employment growth had patterns similar to California’s. We look first at states where housing costs were above the median value across all states, averaged over the years of the study. These high-housing-cost states had 0.32 percentage point slower economic growth per year. But job growth in those same states was slower by nearly the same amount, 0.34 percentage point per year. Since these states had no relative gap in the growth rates of output and jobs, the comparison suggests that higher housing costs may not be the underlying cause of the discrepancy in California.

By contrast, states with high growth in the share of high-wage jobs show a pattern similar to California’s. For states that ranked above the median in high-wage job growth, economic growth increased 0.50 percentage point more per year than in the states below the median. At the same time, overall job growth was only about 0.20 percentage point more per year in the states with high growth of high-wage jobs. This 0.30 percentage point difference between the annual rates of output and job growth can become substantial over many years, adding up to a 6.2% difference over 20 years. This evidence supports the idea that faster high-wage job growth may be an underlying cause of California’s gap between output growth and job growth.

We also conducted statistical tests across states of the relationships between high housing costs and fast high-wage job growth on the one hand and the gap between output growth and job growth on the other hand. We find that housing costs have negative correlations with both per capita GSP growth and job growth. However, these two correlations are extremely close, indicating little correlation between housing costs and the difference between output and job growth. This suggests that high housing costs cannot explain gaps between state output growth and job growth. By contrast, growth in the share of high-wage jobs has a much stronger relationship with a state’s output growth than its job growth. The large difference between these correlations translates into a significant relationship between the growth in the share of high-wage jobs in a state and the gap between output growth and job growth. This suggests that faster high-wage job growth may help explain California’s gap between growth in output and jobs.

Computing these relationships for all states does not necessarily explain California’s experience. Nonetheless, our results are consistent with the idea that California’s economic growth outstripped its job growth because of relatively high gains in the share of high-wage employment. To be sure, this explanation holds for the entire sample period, but does not explain the pattern from 2001 to 2011, when California was below the median in growth of high-wage employment share.

Conclusion

Evidence suggests that the reason California has experienced faster economic growth than job growth is that employment has shifted to high-wage industries. Slower job growth, particularly in low-wage industries, is a potentially important problem if it implies fewer opportunities for less-skilled workers.

A related concern is the growth in the poverty rate over this same period. California’s poverty rate adjusted for housing costs grew over five percentage points from 1990 to 2011, the third largest increase among all states (see Neumark and Muz 2013). Even excluding the Great Recession, California’s growth in the poverty rate still ranked 13th highest among states. This rise in poverty is consistent with relative declines in job opportunities for less-skilled workers. California’s relatively high economic growth combined with its relatively low job growth may have disadvantaged less-skilled workers, highlighting a key challenge facing policymakers. That is, the greater economic efficiency that helps spur economic growth sometimes comes at the cost of social equity.

David Neumark is Chancellor’s Professor of Economics and Director of the Center for Economics & Public Policy at the University of California, Irvine, and a visiting scholar at the Federal Reserve Bank of San Francisco.

Jennifer Muz is a Ph.D. candidate at the University of California, Irvine.

Reference

Neumark, David, and Jennifer Muz. 2013. How Does California’s Economic Performance Compare to the Other States? San Francisco, CA: Next 10 Foundation.

It is easiest to summarize financial markets research by focusing on the stock market, although the analysis can be applied to other financial markets with little modification. The earliest line of financial economics research that connects directly with current work is on efficient capital markets, developed in the 1960s and 1970s (Fama 1970). The defining claim of efficient capital markets theory was that, if investors process information efficiently and price stocks rationally, then future returns cannot be forecast. That’s because, if returns could reliably be forecast to be abnormally high in the future, investors would buy stocks now. That would cause stock prices to rise immediately, bringing future returns down to a normal level. The reverse would take place if returns could be forecast to be abnormally low. Empirical research appeared to confirm that financial market returns can’t be forecast.

An important piece of evidence appears to contradict stock market efficiency: the observed volatility of stock prices. In an efficient market, stock prices should be considerably less volatile than they actually are in the real world. The inability of efficient markets theory to explain price volatility stems at least partly from the fact that it doesn’t take into account investor risk aversion. Recent research has moved away from the efficient capital markets assumption of risk neutrality, postulating instead that investors are averse to risk.

Unfortunately, formal models incorporating risk aversion produce the degree of volatility seen in the real world only with levels of risk aversion that seem implausibly high. Thus, taking investor risk aversion into account does not satisfactorily explain the volatility of stock price movements.

To understand this better, a recap of financial economics theory is helpful. Early stock market gurus such as Benjamin Graham and David Dodd (1940) held that stocks should be traded based on the relationship between their prices and the discounted value of their expected future dividends. This present value model assumed that the discount rate could be held constant. At first glance, the efficient markets model appeared to conflict with this present value model. However, Samuelson (1965) showed that, far from being contradictory, the present value and efficient markets models were essentially equivalent. If stock prices equal expected future dividends discounted at a constant rate, returns in fact can’t be forecast.

Further, the assumption that stock prices equal expected future dividends independent of the volatility of dividends can be justified only if investor risk aversion is excluded. If investors are risk averse, stock prices will depend on how variable dividends are as well as on their expected levels. By ignoring this effect, market efficiency implicitly treats investors as being risk neutral.

In the 1970s and 1980s, empirical evidence that raised questions about the efficient markets model began to surface. Shiller (1981) and others showed that, under the efficient markets model, stock prices should exhibit the same low volatility as dividends themselves.

By “price volatility,” I mean the overall variability of stock prices over time. It is important to distinguish between this definition and a frequently used alternative definition of volatility as the average variability of stock returns, consisting of dividends plus price changes. To see the difference between price volatility and return volatility, assume unrealistically that investors have information that allows them to predict future dividends into the indefinite future with perfect accuracy. Then the present value model implies that stock returns are equal to the constant discount rate and have zero volatility. At the same time though, stock prices will still go up and down as dividends change, which means they will be volatile.

Shiller’s conclusion was based on the fact that the stock price volatility implied by a given dividends model depends on how much information investors are assumed to have about future dividends. If investors cannot predict future dividend growth at all, they will price stocks at a constant multiple of current dividends. The volatility of stock prices relative to dividends will be zero. On the other hand, if investors have information about future dividends, then stock prices relative to dividends will vary over time. The more information about dividend growth investors have, the greater the average price variation. The extreme case assumes that investors can forecast all future dividends. Therefore, the price volatility associated with complete information is the highest level of volatility that can actually occur. In exercises known as variance bounds tests, Shiller and others found that observed price volatility appeared to exceed this maximum level, contradicting the efficient markets model.

Risk aversion

It is easy to see why the efficient markets model implies low price volatility. The rate of return on stock is defined as the dividend yield plus the rate of capital gain. This relation can be shown to imply that price volatility relative to current dividends is due entirely to the degree that future dividend growth and future returns can be forecast. If investors can forecast variations in dividend growth, they will price stocks at a high or low multiple of current dividends depending on whether dividend growth is expected to be high or low. In fact, dividend growth empirically is nearly impossible to forecast, implying that price volatility can be attributed to it only to a very minor extent.

The efficient markets model implies that future returns can’t be forecast. It follows that the efficient markets model combined with the empirical fact that dividend growth is nearly impossible to forecast implies low stock price volatility. Thus, the degree of stock price volatility seen empirically can only occur if we reject market efficiency and hold that future returns contain a large predictable component. In view of the association of market efficiency with risk neutrality, it follows that the market efficiency assumption of risk neutrality must be rejected to allow for risk aversion.

Economists have a measure of risk aversion. Suppose an investor were offered the prospect of either doubling his wealth or halving it, depending on the outcome of a random event. The higher the probability of success, the more prone the investor would be to accept such a gamble. The investor’s risk aversion can be defined to depend on the minimum probability of success that would induce him to accept the bet. The higher the minimum probability he demands, the higher his risk aversion.

An investor with no risk aversion would accept the gamble if the probability of success exceeded one-third because the gain under success is double the loss under failure. But risk-averse investors would require a higher probability of success. An investor who requires a probability of 0.5 is defined as having risk aversion of one. Table 1 shows other probabilities. A reasonable guess for the average investor’s minimum probability is around 0.67, implying a risk aversion of about 2.

Table 1
Probability of success and investors’ risk aversion

In a theoretical model, Lansing and LeRoy (2011) computed the stock price volatility implied by different levels of risk aversion. Like Shiller, they found that, under risk neutrality, predicted maximum stock price volatility is much lower than what is actually seen in the market. They also found that the higher the level of risk aversion, the higher the maximum stock price volatility. To generate a price volatility level near that seen in the real world, the model needs an implausibly high level of risk aversion around 4 or 5, implying a probability of success around 0.94 in the gamble described above. Most investors would not need such a high probability of success to accept the risk.

What’s more, risk aversion must be even higher if the stock price volatility implied by formal models is to be reconciled with actually observed volatility. Maximum volatility in the models is based on the implausible presumption that investors can predict dividends with perfect accuracy into the indefinite future. If investor ability to predict future dividends is more limited, then predicted price volatility will be lower under a given degree of risk aversion. Accordingly, still-higher levels of risk aversion are required to account for real-world price volatility.

To summarize, allowing for risk aversion can in principle generate stock price volatility similar to that seen in real-world financial markets. However, that assumes either that investors are implausibly risk averse or that they can predict dividends into the distant future, or both. Otherwise, price volatility surpasses levels that can be explained by fundamentals. It follows that assuming that investors have reasonable levels of risk aversion is not enough to explain why stock prices are so volatile.

Conclusion

The finance models summarized in this Letter are at odds with empirical data. Economists have been experimenting with innovative models that might help explain why asset prices are so volatile. Some have been looking at behavioral models that don’t assume full rationality, but these have problems of their own. So far, these investigations have not led to convincing explanations. It is far from clear where we go from here.

Stephen F. LeRoy is a professor emeritus at the University of California, Santa Barbara, and a visiting scholar at the Federal Reserve Bank of San Francisco.

References

Fama, Eugene F. 1970. “Efficient Capital Markets: A Review of Theory and Empirical Work.” Journal of Finance 25(3), pp. 383–417.

Graham, Benjamin, and David Dodd. 1940. Security Analysis. New York: McGraw-Hill.

Lansing, Kevin J., and Stephen F. LeRoy. 2011. “Risk Aversion, Investor Information, and Stock Price Volatility.” FRB San Francisco Working Paper 2010-24.

Samuelson, Paul A. 1965. “Proof that Properly Anticipated Prices Fluctuate Randomly.” Industrial Management Review 6, pp. 41–50.

Shiller, Robert J. 1981. “Do Stock Prices Move Too Much to be Justified by Subsequent Changes in Dividends?” American Economic Review 71(3), pp. 421–436.

]]> http://www.frbsf.org/economic-research/publications/economic-letter/2013/april/risk-aversion-stock-price-volatility/feed/ 0 http://www.frbsf.org/economic-research/publications/economic-letter/2013/april/unconventional-monetary-policy-dollar/ http://www.frbsf.org/economic-research/publications/economic-letter/2013/april/unconventional-monetary-policy-dollar/#comments Mon, 01 Apr 2013 07:00:59 +0000 cpatton http://www.frbsf.org/economic-research/?post_type=frbsf_publications&p=28485 After the financial crisis began in 2007, the Federal Reserve reduced the federal funds rate, its main policy tool, close to zero, its lowest possible level. It has remained there since. Because the federal funds rate cannot be reduced further, the Fed has introduced unconventional policy measures to stimulate the economy. One of these unconventional measures is large-scale asset purchases, which are intended to lower long-term interest rates. Another measure is known as forward guidance, communication about the Fed’s expectations for future policy that is intended to guide market expectations and reduce policy uncertainty.

The effectiveness of these new policy tools is an open question. In particular, we don’t know whether the standard channel for transmitting monetary policy through financial markets works as well now as it did in the past. One way to measure the effectiveness of unconventional monetary policy tools is through the U.S. dollar exchange rate. Although the Fed does not target the exchange rate specifically, monetary policy decisions ultimately affect the dollar’s value, which can have important effects on the economy. For example, before the crisis, the dollar typically depreciated following declines in the target for the federal funds rate. The lower value of the dollar in turn helped raise U.S. net exports, boosting output and employment in the United States.

This Economic Letter examines how unconventional policy decisions have affected the value of the. dollar since the Fed lowered the federal funds rate close to zero in December 2008. We look at how the dollar’s value changed during the minutes immediately after Fed policy announcements. This helps isolate the response of the dollar to monetary announcements from other possible factors. In addition, because financial and currency markets may anticipate policy changes and build those expectations into prices, we account for those expectations and focus on the effects of surprise policy announcements.

Our analysis shows that unconventional monetary policy has affected the dollar exchange rate. In particular, surprise unconventional policy easing has pushed down the value of the dollar roughly as much as similar surprise downward moves in the federal funds rate did before the crisis.

Unconventional monetary policy

Identifying how unconventional monetary policy actions have affected the dollar since 2008 is challenging. Because the Fed’s recent actions are unprecedented, we have limited data to work with. To see how unconventional policy actions have affected the dollar’s value, we focus on the dates of monetary policy announcements. We broadly label these quantitative easing announcements, but they could contain news about both large-scale asset purchases and forward guidance. We look at what happened to the trade-weighted value of the dollar measured against a basket of currencies from major U.S. trading partners, including the Canadian dollar, the pound, the euro, and the yen, in a tight time window around these announcements.

The quantitative easing announcements in our sample include statements by the Fed’s policymaking board, the Federal Open Market Committee (FOMC), after scheduled meetings, and speeches and congressional testimony by Fed Chairman Ben Bernanke in which he signaled possible policy changes. Our sample includes all announcements regarding the Fed’s three rounds of quantitative easing.

• The first round began on November 25, 2008, when the Fed announced it intended to buy up to $500 billion in mortgage-backed securities and $100 billion in debt from Fannie Mae, Freddie Mac, and other government-sponsored enterprises.
  

• The second round started with two dates in August 2010: the August 10 FOMC statement announcing that the Fed would roll over its holdings of Treasury securities as they matured, keeping them on the Fed’s balance sheet, and Chairman Bernanke’s August 27 speech at the Economic Symposium in Jackson Hole, Wyoming.

• The third round began with the September 2012 FOMC statement announcing the decision to buy $40 billion in mortgage-backed securities in addition to the ongoing purchases of longer-term Treasuries of $45 billion per month. Another major event in this round was the December 2012 announcement that the Fed expected to wait at least until the economy reached numerical thresholds for unemployment and inflation before it would begin raising the federal funds rate. The FOMC specified these thresholds to help the public understand the Committee’s decisionmaking process and make its forward guidance more precise. (See Glick and Leduc 2013 for a list of announcements used in this study.)
  

We assume policy announcements immediately influence the views of market participants, and that these views are quickly reflected in the value of the dollar. To capture this effect, we look at movements in the trade-weighted value of the dollar in a 30-minute window around each policy announcement in our sample, from 10 minutes before the announcement to 20 minutes after. Using such a narrow time span allows us to isolate policy announcement effects from other possible influences on the dollar’s value. Other studies have used similar currency data (see Neely 2012 and Glick and Leduc 2012), but rely on less-frequent daily data or consider only the first round of quantitative easing.

Surprise policy announcements

How much an announcement affects the dollar’s value depends largely on whether market participants expect it or were surprised by it. If market participants anticipate the news, then no additional information is revealed and the exchange rate should not move. This makes determining market expectations crucial for our analysis.

Before the crisis, when the federal funds rate was the main monetary policy tool, researchers could easily determine the market’s policy expectations by looking at federal funds rate futures contracts. Those futures show the value of the federal funds rate that market participants expect at some future date (see, for instance, Kuttner 2001). But, the federal funds rate is now near zero and is no longer the main monetary policy tool. Thus, federal funds futures don’t tell us much now about the expectations of market participants for unconventional monetary policy.

Quantitative easing is designed to lower longer-term interest rates. That suggests that a potential way of measuring the extent to which market participants expected or were surprised by unconventional policy announcements is to look at long-term Treasury rate futures. Specifically, for any given quantitative easing announcement, we can measure changes in long-term Treasury rate futures over the same 30-minute window used to calculate the change in the dollar’s value. Big swings in futures prices at the time of the announcement suggest that the change in policy surprised participants (for more details, see Wright 2011).

Policy surprises and the dollar

Figure 1
Dollar’s response to quantitative easing surprises

Figure 1 plots the relationship between quantitative easing policy surprises and the trade-weighted value of the U.S. dollar. The chart’s horizontal axis shows the extent of the policy surprise in percentage points. A higher value implies a larger degree of surprise easing in a policy announcement. The vertical axis shows the change in the dollar’s value, also in percentage points. For this measure, a higher value implies a greater dollar appreciation; a negative value implies depreciation.

The figure includes quantitative easing policy surprises with negative values, indicating unexpected policy tightening, and positive values, indicating unexpected policy easing. The largest positive surprises came on January 18, 2008, and March 18, 2009, during the first round of quantitative easing. The figure shows a clear negative relationship between the magnitude of surprise easing and the value of the dollar, as captured by the downward sloping line. In other words, the greater the surprise, the more the dollar depreciates. In fact, the line indicates that a 1 percentage point easing in long-term Treasury futures rates, suggesting a policy surprise, leads within 30 minutes to a roughly 3 percentage point decline in the trade-weighted value of the dollar.

How strong are these effects on the dollar’s value compared with the impact of conventional monetary policy? To gauge this, we compare these results with the effects of surprise changes in the federal funds rate before the financial crisis. This conventional policy sample period runs from January 1994, when the FOMC began issuing press releases whenever it met or changed monetary policy, until October 2008, just before the Committee lowered the federal funds rate close to zero.

Figure 2
Dollar’s response to fed funds rate surprises

We identify surprise changes in monetary policy during this period by examining changes in federal funds rate futures in the same 30-minute window around monetary policy announcements. Figure 2 shows that the dollar tended to depreciate following surprise federal funds rate easing, measured along the horizontal axis. A 1 percentage point surprise in the federal funds rate causes the dollar to drop about 0.7 percentage point.

However, a surprise change in the Fed’s target for the federal funds rate is different from a surprise change in quantitative easing. The federal funds rate is an overnight interest rate, while quantitative easing involves longer-term securities. For instance, if the FOMC wanted to move the federal funds rate to engineer a quarter percentage point fall in the 10-year Treasury rate, it would typically have to target a decline in the federal funds much larger than a quarter percentage point.

To make an apples-to-apples comparison on how conventional and unconventional policy surprises affect the dollar, we need to develop a way of making the two types of policy announcements equivalent for measurement purposes. To do that, we look at how long-term interest rate futures changed on average following a surprise change in the federal funds rate during our pre-crisis sample period. We use this estimate to translate our quantitative easing surprises into an equivalent measure to compare with federal funds rate surprises (see Glick and Leduc 2013 for details).

We find that a quantitative easing surprise equivalent to a 1 percentage point decrease in federal funds rate futures leads to a 0.5 percentage point depreciation in the dollar. The size of this effect is comparable with the 0.7 percentage point depreciation following surprise movements in the federal funds rate before the financial crisis.

Conclusion

Our study shows that unconventional monetary policy has affected the value of the dollar. Moreover, changes in the dollar’s value immediately following surprise policy announcements are comparable before and after the crisis. This suggests that changes in unconventional monetary policy have affected the dollar about as much as changes in the federal funds rate did before the financial crisis.

It is more difficult to assess whether these changes in the dollar’s value stemming from unconventional monetary policy have similar effects on U.S. net exports as those stemming from conventional policy. The recent boost to net exports from a weaker dollar may have been obscured by other factors, such as reductions in foreign demand stemming from uncertainty about Europe’s economic recovery.

Reuven Glick is a group vice president in the Economic Research Department of the Federal Reserve Bank of San Francisco.

Sylvain Leduc is a vice president in the Economic Research Department of the Federal Reserve Bank of San Francisco.

References

Glick, Reuven and Sylvain Leduc. 2013. “The Effects of Unconventional and Conventional U.S. Monetary Policy on the Dollar.” Manuscript, Federal Reserve Bank of San Francisco.

Glick, Reuven, and Sylvain Leduc. 2012. “Central Bank Announcements of Asset Purchases and the Impact on Global Financial and Commodity Markets.” Journal of International Money and Finance 31(8), pp. 2,078–2,101.

Kuttner, Kenneth. 2001. “Monetary Policy Surprises and Interest Rates: Evidence from the Fed Funds Futures Market.” Journal of Monetary Economics 47(3), pp. 523­–544.

Neely, Christopher. 2012. “The Large-Scale Asset Purchases Had Large International Effects.” FRB St. Louis Working Paper 2010-018.

Wright, Jonathan. 2011. “What Does Monetary Policy Do to Long-Term Interest Rates at the Zero Lower Bound?” National Bureau of Economic Research Working Paper 17154.

This Economic Letter explores whether the recent pattern in China’s official GDP data is consistent with a range of alternative indicators of Chinese economic activity. Some of these indicators are produced by agencies of the Chinese government. We also examine trading-partner data on exports to and imports from China produced by non-Chinese sources. Historically, both these domestic and foreign indicators have closely tracked official Chinese output statistics. In 2012, they were broadly consistent with China’s reported slowdown in growth of industrial production and GDP. Moreover, these data are consistent with the pickup in growth observed in the fourth quarter of 2012 from earlier in the year. Hence, we find no evidence that China’s slowdown in 2012 was greater than officially reported.

Evidence from domestic Chinese data

Since the 1990s, a number of commentators have expressed concern about the accuracy of Chinese statistics. The challenges of producing accurate statistics are substantial in an economy growing as fast as China’s, in which the structures of production and expenditure are changing rapidly (see Holz 2008). Furthermore, even prominent Chinese political leaders have expressed concern that certain key statistics might sometimes be manipulated for political advantage. For example, in remarks to the U.S. ambassador in 2007, Vice Premier Li Keqiang stated, “GDP figures are ‘man-made’ and therefore unreliable,” according to a Wikileaks release of a U.S. diplomatic cable (see Wikileaks 2007).

Figure 1
Recent GDP figures consistent with other indicators

China’s GDP data move closely in sync with other high-profile official indicators, such as industrial production and retail sales. However, these indicators might themselves be distorted. In addition, to the extent these indicators are incorporated directly into GDP as inputs, they might not vary independently.

To test the accuracy of Chinese GDP data, we compared them with a range of alternative domestic indicators available over a long time span that seem less subject to manipulation. We grouped these alternative indicators into two sets. The first set, labeled “Li” in Figure 1, was based on the preferences of Vice Premier Li, as reported in the 2007 U.S. diplomatic cable. He said he got a better read on the pace of economic activity by looking at electricity production, rail cargo shipments, and loan disbursements. In his view, those data were relatively accurate. The second set of alternative indicators, labeled “Broad” in Figure 1, included an index of consumer sentiment, construction of new floor space, an index of raw materials usage, air passenger volume, and the nominal value of new residential real estate construction.

We then examined year-over-year values for GDP and our two sets of alternative indicators. For each of the sets, we performed a statistical exercise known as principal components. This method synthesized the fluctuations in these indicators into two indexes, one for each of the two sets. Creating summary indexes rather than using the eight indicators individually simplified the analysis.

Next, we looked at the statistical relationship between the two alternative indicator indexes and China’s GDP over a sample period from the first quarter of 2000 through the third quarter of 2009. We then used the information on the statistical relationship between GDP and the indexes during this period to project GDP growth from the fourth quarter of 2009 through 2012. This allowed us to check whether recent data are consistent with the relationship observed from 2000 through 2009. For details concerning our techniques, see our technical appendix at http://www.frbsf.org/publications/economics/letter/2013/el2013-08-technical-appendix.pdf. Data used in the study are available at http://www.frbsf.org/publications/economics/letter/2013/el2013-08-data-appendix.xlsx.

Figure 1 plots the relationship between reported GDP growth and our two alternative indicator indexes. It includes the statistical predictions made for 2009 through 2012, outside the sample period. We do not expect a precise match between our projections and GDP as actually reported because the predictions are based on an approximate statistical model subject to various sources of error.

For the third quarter of 2012, this method predicts values for year-over-year GDP growth that represent a slightly larger slowdown than China officially reported. By contrast, for the fourth quarter of 2012, both alternative indicator indexes turned less negative and the predicted values exceed officially reported GDP. The recent misses are not large in statistical terms.

Our statistical method tells whether recent data are consistent with the relationship observed in the past. We conclude that the typical relationship between the alternative indicators and GDP observed from 2000 to 2009 continued to hold through 2012. In other words, the 2012 slowdown in reported growth—and its modest rebound in the fourth quarter—do not deviate from previously reported patterns. However, our results don’t provide evidence on the overall, long-term accuracy of Chinese GDP statistics. For example, if Chinese GDP had been consistently overreported over time, then our measured statistical relationships would be skewed by this persistent bias.

Consistency with data reported outside China

While neither of our alternative indicator indexes seems likely to have been substantially manipulated, all the components were produced by Chinese authorities. As such, our exercises may only partially address concerns about the accuracy of Chinese data. Therefore, we also compare official Chinese GDP figures with external indicators of Chinese economic activity.

Specifically, we look at Chinese imports and exports reported by its trading partners. Exports to China are likely to be positively correlated with Chinese aggregate demand. Moreover, to the extent that these exports are intermediate inputs for Chinese manufacturing, they would also be positively correlated with industrial production. In turn, imports of Chinese products generally reflect external demand for Chinese goods and services.

We used two sources of trading-partner data. The first is based on trade with the United States, the European Union, and Japan, a group we call the “trio.” These three trading partners account for about half of global imports from China and about a third of exports to China. Trade data from these partners are available with a relatively short lag.

The second trade-data source is based on China’s overall global imports and exports. The International Monetary Fund collects these data from individual countries, but they are available with a longer lag than data from the trio. In all cases, we include Hong Kong in our measures because it serves as an intermediary for Chinese trade. It is often hard to differentiate between trade with Hong Kong and with the rest of China for statistical purposes.

Figure 2
Chinese GDP consistent with trading-partner data

We transformed the data in three ways. First, the trade figures are in U.S. dollars, so we adjusted exports and imports for inflation by using U.S. price data for all countries. For exports, we constructed a price index from commodity-specific export price indexes weighted by the nominal commodity shares of U.S. exports to China. For imports, we made an inflation adjustment that included all products except petroleum. Statistical results were similar when we used nominal, unadjusted trade values. Second, we used year-over-year growth rates. Third, for the trio on the one hand and the world data on the other hand, we combined export and import growth into two indexes using the principal components method, as we did with the Chinese indicators. Because exports and imports are highly correlated, results turned out to be qualitatively similar whether we used exports and imports individually or combined.

Figure 3
Chinese industrial production also consistent

We found that, in statistical terms, trading-partner data are significantly and positively correlated with Chinese GDP and industrial production. Figure 2 shows the predictions for GDP growth based on trio and overall world exports to China. As in Figure 1, we estimated the relationships based on data through the third quarter of 2009 and predicted GDP for the fourth quarter of 2009 through 2012. The errors are sometimes large. Nonetheless, the slowdown in China’s GDP from 2011 to 2012 is roughly consistent with the values predicted by the trading-partner data.

Figure 3 shows the same exercise using monthly industrial production data. Not surprisingly, the link with trade statistics is even closer because both involve tangible goods. The model does a good job at capturing the 2008 downturn and 2009 resurgence in industrial production. In 2010, China’s industrial production growth was lower than predicted. But the match in 2011 and 2012 is reasonably tight. Hence, the recent Chinese industrial production data also appear to be consistent with Chinese trade volumes as reported by trading partners. The fit is even closer when only China’s imports are used in the exercise.

Conclusion

We found that reported Chinese output data are systematically related to alternative indicators of Chinese economic activity. These include alternative indicator indexes of Chinese activity composed of variables that are less susceptible to official manipulation, as well as externally reported trade volume measures. Importantly, these models suggest that Chinese growth has been in the ballpark of what official data have reported. We find no evidence that recently reported Chinese GDP figures are less reliable than usual.

John Fernald is a senior research advisor in the Economic Research Department of the Federal Reserve Bank of San Francisco.

Israel Malkin is a research associate in the Economic Research Department of the Federal Reserve Bank of San Francisco.

Mark Spiegel is a vice president in the Economic Research Department of the Federal Reserve Bank of San Francisco.

References

Bradsher, Keith. 2012. "China Data Mask Depth of Slowdown, Executives Say.” New York Times, June 22.

Holz, Carsten A. 2008. "China’s 2004 Economic Census and 2006 Benchmark Revision of GDP Statistics: More Questions than Answers?" The China Quarterly, March.

Wikileaks. 2007. “Fifth Generation Star Li Keqiang Discusses Domestic Challenges, Trade Relations with Ambassador.” Cable, March 15. http://wikileaks.org/cable/2007/03/07BEIJING1760.html

This Economic Letter summarizes recent research (Dunn, Liebman, and Shapiro 2012a, b, c) that pinpoints the distinct sources of medical-care spending growth in the employer-sponsored and private health insurance market. The privately insured health-care market is economically important. Total spending for employer-sponsored private health insurance was $709 billion in 2011 (Gaynor and Newman 2012), which was approximately 30% more than Medicare outlays that year. Unlike the Medicare market in which CMS fixes payments to providers, private-sector prices are set through negotiations between insurers and providers. As a result, those prices are sensitive to competitive factors. Thus, spending growth in the privately insured market can stem from a multitude of sources, including growth in negotiated services prices.

Identifying the components of medical spending growth

From 2003 to 2007, medical-care expenditures per enrollee grew 28%, an average of 6.3% annually, faster than the 20% growth in nominal per capita GDP. Spending may have grown so rapidly relative to GDP for a few reasons. The growth-accounting framework of Dunn, Liebman, and Shapiro (2012a, b, c) tracks four potential sources of this growth: One, the prices of procedures may be outpacing general price inflation. Two, patients may be receiving a higher quantity of services for treatment. Three, a higher percentage of enrollees may be seeking medical care. Four, the population of insured individuals may be aging.

Dunn, Liebman, and Shapiro (2012a, b, c) assess these components by analyzing retrospective claims data contained in the Thomson Reuters MarketScan research database for a sample of employer-sponsored and privately insured patients. These data provide payment information from employer and health-plan sources, and contain medical and drug data for several million commercially insured individuals, including employees, spouses, and dependents. The data were then processed using the Ingenix Symmetry program, which assigns each claim to a particular condition category and episode of care. Specifically, illnesses, injuries, and other conditions requiring medical care are classified by condition groupings and severity. The episode of care for a particular patient with a specific condition begins at the time of initial treatment and ends at final treatment.

Figure 1
Breakdown of medical-care expenditures

Figure 1 displays a breakdown of medical-care expenditures. Spending is initially assessed on a per capita basis and adjusted using demographic population weights that control for changes in age and gender distribution. Spending after applying the weights is labeled demographically adjusted spending per enrollee. Because demographically adjusted spending does not equal actual spending, a difference exists. This remaining portion is labeled the demographic residual.

Demographically adjusted spending per enrollee can be divided between expenditure per episode of treatment and episodes per enrollee. The latter is a measure of the prevalence of treatment for a condition. For example, in the case of hypertension, we track the number of episodes of treatment for hypertension per enrollee as well as the expenditure per episode of treatment for this condition. Finally, expenditure per episode of treatment is split into service price and service utilization, which is the quantity of services per episode. Service price represents the payment for a specific service, for example, a 15-minute office visit. Service utilization represents the quantity of services performed during an episode of treatment, taking into account the relative intensity of the service. For example, in our methodology, a 30-minute doctor office visit is considered a higher quantity of services than a 15-minute office visit.

Figure 2
Medical-care expenditure growth components, 2003-07

Source: MarketScan Research Database.

Figure 2 shows growth in these individual components of medical-care spending. The index representing spending per enrollee increased 28% between 2003 and 2007, meaning that overall medical-care spending grew approximately 6.3% annually. Episodes per enrollee increased 10.3% over the four-year period, while medical-care prices grew 15.9%, or 3.8% annually. Notably, price growth for medical-care services was not much higher than inflation in the overall economy. The personal consumption expenditure price index, a widely followed inflation measure, rose 11.5% over our sample time frame. Adjusting for inflation, real medical service prices rose only 4% over the period.

Figure 2 also shows that the quantity of services consumed per episode did not change over the sample period. This suggests that substituting more expensive procedures, lengthening office visits or hospital stays, or adding more procedures are not contributing to spending growth. Rather, an increase in the proportion of enrollees receiving treatment is driving most real spending growth. It should be noted that this accounting methodology is not able to determine whether more enrollees are getting treatment because a larger percentage of them are getting ill, more of them are realizing they are sick, or more treatment options are available.

Growth patterns of selected condition categories

To get a better sense of which types of diseases are contributing to growth, Dunn, Liebman, and Shapiro (2012b) report growth patterns for specific condition categories. Each category is calculated as a weighted average of the many underlying condition indexes in that category. Condition indexes track growth in prices, utilization, and prevalence at the condition level. Figure 3 breaks down growth between 2003 and 2007 for the four largest condition categories: orthopedics, cardiology, gastroenterology, and gynecology. It also analyzes conditions grouped into the preventive-care category, which includes routine checkups. These five categories represented approximately half of all spending in 2003.

Figure 3
Nominal medical-care expenditure growth, 2003-07

Source: MarketScan Research Database.

Note: Numbers in parentheses show disease category shares of the overall spending level.

Figure 3 shows two interesting results. First, the quantity of services per cardiology episode fell 7% between 2003 and 2007, a sizeable decrease. Second, spending for preventive care grew tremendously.

The decline in the quantity of cardiology services held down spending growth for cardiology-related treatments. Per capita spending for cardiology conditions grew only 18% over the sample period. A shift in treatment from inpatient to outpatient care largely explains the decline in the quantity of cardiology services performed. Thus, service substitution seems to account for much of the slow growth in spending in this area.

Spending for preventive treatment increased at an extremely rapid rate over the four-year sample period. Growth in the percentage of enrollees getting preventive treatment, that is, episodes per enrollee, drove half this increase. That indicates people are seeking preventive care at an increasing rate. At the same time, medical care generally shifted away from treating late-stage illnesses. For instance, the prevalence of treatment for late-stage ischemic heart disease, late-stage colon cancer, and late-stage breast cancer all decreased. Thus, treatment has shifted towards preventive care and away from care for late-stage illnesses.

Conclusion

An analysis of the components of medical-care expenditures indicates that spending growth in the privately insured market is being driven by the number of treated enrollees as opposed to the cost of treatment. In fact, patterns of utilization of medical services held spending growth in check. This is most evident for cardiology conditions, in which the quantity of services per episode of care declined sizably over the sample period.

Thus, “bending the cost curve” does not necessarily imply reducing growth in the cost of treatment. Rather, it may also imply slowing the growth in the number of enrollees receiving medical treatment. Treatment growth is most pronounced for preventive care. But we are skeptical that holding down growth in this area would be beneficial. In fact, a higher percentage of enrollees receiving preventive treatment may lead to lower expenditures in the future, better health outcomes, or both. Ultimately, more research is needed to determine which forms of spending growth are wasteful and which are productive in terms of health outcomes.

A shift from inpatient to outpatient services has caused utilization of services for certain conditions to decline. At the same time though, some areas, such as cancer treatment, have seen growth in both service utilization and prices. In the case of cancer, we hypothesize that cost growth reflects extensive innovation in treating malignancies. A more comprehensive study of cancer treatment would lead to a better understanding of the rising costs in this area.

Adam Shapiro is an economist in the Economic Research Department of the Federal Reserve Bank of San Francisco.

References

Dunn, Abe, Eli Liebman, and Adam Shapiro. 2012a. “Implications of Utilization Shifts on Medical-Care Measurement.” Bureau of Economic Analysis Working Paper.

Dunn, Abe, Eli Liebman, and Adam Shapiro. 2012b. “Decomposing Medical-Care Expenditure Growth.” Federal Reserve Bank of San Francisco Working Paper 2012-26.

Dunn, Abe, Eli Liebman, and Adam Shapiro. 2012c. “Developing a Framework for Decomposing Medical-Care Expenditure Growth: Exploring Issues of Representativeness.” Forthcoming in Measuring Economic Sustainability and Progress, NBER Book Series Studies in Income and Wealth, eds. Dale Jorgenson, Steven Landefeld, and Paul Schreyer.

Gaynor, Martin, and David Newman. 2010. Health Care Cost and Utilization Report: 2010. Washington, DC: The Health Care Cost Institute.

Keehan Sean P., Andrea M. Sisko, Christopher J. Truffer, John A. Poisal, Gigi A. Cuckler, Andrew J. Madison, Joseph M. Lizonitz, and Sheila D. Smith. 2011. “National Health Spending Projections through 2020: Economic Recovery and Reform Drive Faster Spending Growth.” Health Affairs 30(8), pp. 1,594–1,605.