Recent trends in income and wealth inequality have drawn much attention from both academics and the general public. Atkinson, Piketty, and Saez (2011) and Saez and Zucman (2014), among others, document these trends for different countries around the world and have prompted a substantive debate about the underlying causes and the appropriate policy responses, if any.

*A General Approach*

In new research (Fernholz, 2015), I develop a general statistical model of wealth distribution to address these issues empirically. This approach overlaps both the empirical literature on income dynamics (Guvenen, 2009; Browning et al., 2010) and the broader literature on Pareto distributions for income, wealth, and other economic variables (Gabaix, 1999; Benhabib et al., 2011; Jones and Kim, 2014). A key difference between my approach and these literatures, however, is that I impose no parametric structure on the underlying processes of household wealth accumulation and do not model or estimate these processes directly.

Despite this minimal structure, I use new techniques to obtain a simple household-by-household characterization of the stable distribution of wealth. Using this characterization, I show how the model can replicate any empirical distribution and then construct such a match for the 2012 U.S. wealth distribution as described by the data of Saez and Zucman (2014).

*The U.S. Wealth Distribution, Present and Future*

The U.S. wealth shares data of Saez and Zucman (2014) have attracted attention recently by demonstrating an upward trend in top wealth shares starting in the 1980s, an observation that suggests that U.S. wealth is transitioning from one distribution to another. If this is the case, it is natural to wonder where the distribution is going.

My statistical approach to wealth distribution is able to provide estimates of the future U.S. stable distribution. These estimates depend only on the current rate of change of wealth shares in the economy, not on the causes of these changes. I consider four transitioning wealth shares scenarios:

- The 2012 U.S. wealth distribution is stable.

- The share of wealth held by the top 0.01% of households is increasing by 1% per year, while the share of wealth held by the bottom 90% of households is decreasing by 0.5% per year.

- The shares of wealth held by the top 0.01% and 0.01-0.1% of households are increasing by 1.5% and 0.5% per year, respectively, while the share of wealth held by the bottom 90% of households is decreasing by 1% per year.

- The shares of wealth held by the top 0.01% and 0.01-0.1% of households are increasing by 3% and 1% per year, respectively, while the share of wealth held by the bottom 90% of households is decreasing by 1.5% per year.

In Figure 1, I plot the future stable distributions of wealth for these four scenarios. The substantially different outcomes implied by these scenarios demonstrate how sensitive the future distribution is to upward trends in today’s top wealth shares. In the case of Scenario 4, top wealth shares are increasing so quickly that the distribution never stabilizes but instead separates into divergent subpopulations, with the top 0.01% of households eventually holding all wealth (represented by the vertical line in Figure 1). I emphasize that these hypothetical estimates are not intended as predictions but rather as descriptions of the trajectory of inequality in the absence of changes in the economic environment.

**Figure 1:** Household wealth shares under Scenarios 1 (solid black line), 2 (dashed red line), 3 (dotted blue line), and 4 (vertical dot-dashed green line).

Interestingly, it is difficult to reject a divergent trajectory as in Scenario 4 for the current U.S. wealth distribution. Indeed, according to the data of Saez and Zucman (2014), the share of wealth held by the top 0.01% of households in the U.S. has increased by an average of more than 3.5% per year since 2000 and more than 4% per year since 1980—faster than is assumed in Scenario 4. A cautious interpretation of this fact is that further increases in top wealth shares are likely in the near future.

*Estimating the Effects of Progressive Capital Taxes*

Ever since Piketty (2014) proposed a progressive capital tax in response to increasing wealth inequality, much of the debate surrounding this policy has centered on how such taxes are likely to increase government revenues or distort economic outcomes, rather than how they might affect the distribution of wealth. My model can provide estimates of the distributional effects of progressive capital taxes on the U.S. economy. These purely empirical estimates do not rely on any assumptions about the underlying causes of inequality.

Because it is a useful baseline case to consider, the estimates I present here assume that a capital tax rate of 1% on some subset of households reduces the growth rate of wealth for those households by 1%. Although this leaves out incentive effects of the taxes, these effects can be incorporated into the model by adjusting the tax’s impact on the growth rates of household wealth. In Figure 2, I show the effects of a progressive capital tax similar to the policy proposed by Piketty (2014). In particular, this tax sets the capital tax rate for the top 0.5% of U.S. households equal to 2% and the rate for the top 0.5-1% of U.S. households equal to 1%, while the remaining 99% of households are assumed to neither pay nor receive any tax or subsidy.

Figure 2 displays the effects of this tax on the U.S. wealth distribution, assuming that the 2012 U.S. wealth distribution is stable (Scenario 1). As the figure shows, a progressive capital tax imposed on just 1% of households in the economy reshapes the distribution of wealth and reduces inequality. In fact, the after-tax stable distribution of wealth in this case is similar to the distribution observed in the U.S. in 1978 (the dotted green line in Figure 2), one of the most egalitarian in the U.S. in the last century.

**Figure 2:** Household wealth shares with (dashed red line) and without (solid black line) a 1-2% progressive capital tax on the top 1% of households under Scenario 1, and for the U.S. in 1978 (dotted green line).

I emphasize that this result is neither a statement about total welfare nor an endorsement of a progressive capital tax. My statistical approach only generates empirical estimates of the distributional effects of taxes and other policies, it does not measure the potentially large distortions or costs associated with such policies.

*Interpretation*

A measured interpretation of my results is important. Indeed, there remains uncertainty about the true state of U.S. wealth inequality today. While the quantitative results about the future U.S. wealth distribution and the effects of progressive capital taxes shown in Figures 1-2 are useful, a more important contribution is to introduce an empirical methodology that can address these questions without making assumptions about the causes or consequences of inequality. As more wealth shares data become available in the future, this methodology should yield even more information about the future of U.S. wealth inequality.

*References*

Atkinson, A. B., T. Piketty, and E. Saez (2011, March). Top incomes in the long run of history. Journal of Economic Literature 49(1), 3-71.

Benhabib, J., A. Bisin, and S. Zhu (2011, January). The distribution of wealth and fiscal policy in economies with infinitely lived agents. Econometrica 79(1), 123-157.

Browning, M., M. Ejrnæs, and J. Alvarez (2010, October). Modelling income processes with lots of heterogeneity. Review of Economic Studies 77(4), 1353-1381.

Fernholz, R. T. (2015, January). A Statistical Model of Inequality. mimeo, Claremont McKenna College.

Gabaix, X. (1999, August). Zipf’s law for cities: An explanation. Quarterly Journal of Economics 114(3), 739-767.

Guvenen, F. (2009, January). An empirical investigation of labor income processes. Review of Economic Dynamics 12(1), 58-79.

Jones, C. I. and J. Kim (2014, October). A Schumpeterian model of top income inequality. mimeo, Stanford GSB.

Piketty, T. (2014). Capital in the Twenty-First Century. Cambridge, MA: Harvard University Press.

Saez, E. and G. Zucman (2014, October). Wealth inequality in the united states since 1913: Evidence from capitalized income tax data. NBER Working Paper 20265.

*This post written by Ricardo Fernholz*.

Just after Republican lawmakers announced they were rejecting raises for rank-and-file state troopers, Gov. Scott Walker’s administration granted $4-an-hour raises to the State Patrol officers

responsible for protecting the governor.…

The pay bump will cost the state about $36,500 and comes at a time

when the Republican governor is routinely traveling out of state — with the officers for security — to attend what amount to campaign events leading up to a presidential run.The top two Republicans in the Legislature announced Feb. 17 they would not allow a 17% pay boost for state troopers that Walker’s

administration had negotiated. Those pay raises can’t go into effect without the approval of lawmakers.A separate pay raise took effect five days later for the troopers belonging to the State Patrol’s Dignitary Protection Unit who are responsible for protecting the governor 24 hours a day, State Patrol spokeswoman Peg Schmitt said Monday. That pay increase does not require legislative approval.

WKOW-TV in Madison first reported on the raise, which reportedly will go to 10 troopers.

These funds will come from the Department of Transportation. The WKOW report observes that “DOT officials estimate they are facing a budget shortfall of $680 million heading into the 2015-17 biennium.”

]]>For the project we assembled annual data on the interest rate set by the central bank (or close substitute) along with inflation estimates for 20 different countries going back in some cases to 1800, along with more detailed quarterly data since 1970. We constructed a proxy for expected inflation using autoregressions estimated over rolling windows. The figure below plots the resulting annual series for the ex-ante real interest rates for eight of the countries we looked at.

We found little support in these data for two of the popular conceptions many people have about real interest rates. First, although it is often assumed in theoretical models that there is some long-run constant value toward which the real interest rate eventually returns, our long-run data lead us to reject that hypothesis, consistent with other studies of postwar data by Caporale and Grier (2000), Bai and Perron (2003), and Rapach and Wohar (2005). For most countries, real rates were significantly higher before World War I than they have been since. The world wars were associated with big negative real rates, spectacularly so for countries like Germany, Japan, Italy, and Finland (not shown in the figure above).

We also found little support for the popular assumption that the long-run economic growth rate is the primary factor driving changes in the equilibrium real interest rate over time. The graph below summarizes the U.S. experience over the last 7 economic expansions. The figure plots the average annual real GDP growth rate from business cycle peak to peak on the vertical axis and the average real interest rate on the horizontal axis. The correlation is actually negative, contrary to the popular assumption that a higher long-run GDP growth rate would be associated with a higher long-run equilibrium real rate, though we could get a modestly positive correlation if we threw out the brief expansion ending in 1981:Q3.

Using the longer annual data set, we found a positive correlation between the growth rate and real interest rate across business cycles, (correlation of +0.23 and R^{2} of 0.05). But this correlation would become -0.23 if we dropped the cycles ending in 1920 and 1948.

If we look at cross-sectional evidence of 30-year averages of growth rates and real rates across countries, we find a correlation of 0.42 (R^{2} = 0.18). But this correlation would become negative if Australia had not been included. We come away from our investigation quite skeptical of any analysis that puts growth of actual or potential output at the center of long-run real interest rate determination.

Our paper provides a detailed narrative discussion of what seems to explain changes in long-term averages of the real interest rate over time. We conclude that changes in personal discount rates, financial regulation, trends in inflation, bubbles and cyclical headwinds have had important effects on the average real rate observed over any given decade. We examine the secular stagnation hypothesis in detail. On balance, we find it unpersuasive, concluding that it confuses a delayed recovery with chronically weak aggregate demand.

One point that is often lost in discussions of these issues is the role of the monetary tightening cycle. As seen in the figure below, the gap between the nominal fed funds rate and the trailing one-year core inflation rate became quite high at the peak of each of the previous 5 monetary tightening cycles.

Nevertheless, if one were to look at the size of that gap at the point when output equaled potential GDP– a point we have yet to reach in the current cycle– the real interest rate at that point would have still appeared to be quite low in several of those episodes.

It’s worth remembering that recoveries from financial crises often take many years. Based on indicators such as housing investment as a percent of GDP, there is still significant potential for cyclical expansion in the U.S. Our paper reviews a great deal of evidence that leads us to conclude that those who see the current situation as a long-term condition for the United States are simply over-weighting the most recent data from an economic recovery that is still far from complete.

We found what appears to be a stable statistical relation in our long-run data set in the form of a cointegrating relation between the U.S. ex-ante real interest rate (plotted in black in the figure below) and a measure that is similar to the median of 30-year-moving average real rates across the world (shown in blue). When the U.S. is below the long-run world rate, as it is now, we’d expect the U.S. rate to rise.

We also calculated the current forecast implied by that cointegrating relation, which is shown in the figure below. We can have confidence that the U.S. and world rates will eventually converge to each other, but the exact value to which they will converge is subject to growing uncertainty the farther into the future we try to project, as a necessary consequence of the permanent and sometimes dramatic shocks to world equilibrium real rates that characterize the historical data.

Finally, our paper discusses the implications of these findings for monetary policy. Orphanides and Williams have noted that if the central bank does not know the true value of long-run equilibrium magnitudes like the real interest rate, it pays to incorporate more inertia in the conduct of monetary policy. We perform some simulations using the FRB/US model to gauge the relevance of this concern in the current setting. We conclude that, given that we do not know the equilibrium real rate, there may be benefits to waiting to raise the nominal rate until we actually see some evidence of labor market pressure and increases in inflation. Relative to the “shallow glide path” for the funds rate that has featured prominently in recent Fed communications, our findings suggest that the funds rate should start to rise later but– provided the recovery does gather pace and inflation picks up– somewhat more steeply.

]]>

They are log GDP for the first terms of the most recent seven presidencies, SAAR, normalized to the quarter of inauguration.

**Figure 2:** Nominal GDP, billions of $, SAAR, normalized to quarter of inauguration of first term for Obama (blue), GW Bush (red), Clinton (green), GHW Bush (black), Reagan (teal), Carter (purple), Nixon (chartreuse). Source: BEA (2014Q4 second release), and author’s calculations.

Some would object that using nominal GDP is misleading. In defense of reporting nominal magnitudes, reader Ironman writes:

It is our practice to always present nominal data because it is the data that doesn’t change as a result of inflation adjustments, which are always arbitrary in practice and are always in need of being updated, since almost all readers prefer that kind of information to be presented in terms of constant, current day dollars. Since we provide the relevant links to all original data sources, anyone who wants to confirm our numbers can get them and not wonder how they have been adjusted. Our readers are a pretty sharp bunch and are pretty capable of adjusting the nominal data to account for whatever measure of inflation they might like to consider, whether CPI-U, GDP deflator, etc.

Here for comparison is real GDP.

**Figure 3:** Real GDP, billions of Ch.2009$, SAAR, normalized to quarter of inauguration of first term for Obama (blue), GW Bush (red), Clinton (green), GHW Bush (black), Reagan (teal), Carter (purple), Nixon (chartreuse). Source: BEA (2014Q4 second release), and author’s calculations.

If you can do deflation, and the relevant graphing, in your head…you are a better man than I!

My view: deflation into real terms, even if not perfectly done, makes sense. The argument that deflators are revised over time, and hence makes deflation problematic is interesting, but in my mind not dispositive, given that nominal series are revised over time as well. For instance, after the third release of the 2014Q4 GDP figures next month, the series will be once again revised in the annual benchmark. Years thereafter, the real and nominal series will be revised again, as more data becomes available. So, unless the question pertains to nominal magnitudes instead of real (or one believes the deflators to be particularly problematic), it makes sense to report the variables in…real terms.

Here’s the real GDP series, for the entire presidential terms.

**Figure 4:** Real GDP, billions of Ch.2009$, SAAR, normalized to quarter of inauguration of first term for Obama (blue), GW Bush (red), Clinton (green), GHW Bush (black), Reagan (teal), Carter (purple), Nixon (chartreuse). Source: BEA (2014Q4 second release), and author’s calculations.

Normally, we’re entertained by Chinn’s analysis, since it frequently involves comparisons of the job growth between Wisconsin and its western neighbor Minnesota since Walker was sworn into office in January 2011, which we find funny because of all the states surrounding Wisconsin, the composition of Wisconsin’s economy is much less similar to Minnesota than it is to any of the states with which the state shares waterfront footage on Lake Michigan, which is something that one might think an economics professor at the University of Wisconsin-Madison would know.

[Update, 3/1] I have calculated a similarity index for MN, IL, MI vs. WI, based on output composition. It’s an unweighted measure of absolute sector differences, (Σ|x_{WI}-x_{i}|)/n . The indices are 0.010, 0.013, 0.007, for MN, IL, MI, respectively. In other words, MI is the most similar to WI, MN next. And, interestingly, MI far outpaces WI, in Figure 1 below.

Gee, I think I’d done this comparison, somewhere in the dim past. Oh, it was August 20, 2014, a full five months ago. In Figure 2 of that post, Wisconsin lagged all her neighbors.

Perhaps the choice of states was in dispute. I used adjoining states; Political Calculations appears to favor the region defined by this map (Census region Great Lakes).

**Source:** Political Calculations (February 26, 2015).

So, let’s examine the relative performance of Wisconsin against the neighbors, defined by Political Calculations. (Side note: Ohio does not have waterfront on Lake Michigan.)

**Figure 1:** Log coincident indices for Wisconsin (bold red), Minnesota (blue), Illinois (green), Michigan (teal), Indiana (purple), Ohio (chartreuse), United States (black), all normalized to 2011M01=0, seasonally adjusted. Numbers on right hand side (color coded) refer to log-differences relative to NBER defined peak of 2007M12. Source: Philadelphia Fed, and author’s calculations.

If my eyes do not deceive, the bold red line (Wisconsin) lies below all other series. Had I included Kansas, well, you know where that state’s index would lie. (Hint: as of December, it is 1.5 percentage points below that of Wisconsin’s.) As indicated in the notes to Figure 1, the Philadelphia Fed data are readily available for download in Excel spreadsheet should one want to check my calculations.

More on other indicators, from Flavelle/Bloomberg View.

Oh, here is the latest on the Quarterly Census on Employment and Wages, which astute readers will recall (e.g., here) is the series that Governor Walker was for before he was against: Wisconsin’s job creation remained sluggish in latest 12-month report:

Turning in another laggard job-creation report, Wisconsin gained 27,489 private-sector jobs in the 12 months from September 2013 to September 2014, according to data released Tuesday by the state Department of Workforce Development.

The QCEW is a *census* while the BLS household survey series that Political Calculations displays in this graph is based on a survey (I think; the graph identifies the source as Census, while I have an identical series that is sourced from BLS). That is why the Walker Administration originally favored the QCEW over the establishment survey (Apparently that viewpoint is now “inoperative”, since I no longer hear the QCEW lauded by Walker administration officials.)

So, in summary, the data I have indicate that Wisconsin’s economic performance since 2011M01 has been lackluster. I would welcome actual data indicating otherwise. (Side question: Why are almost every series plotted in the Political Calculations post reporting in nominal terms? The sole exception is tax collections per employee, which would seem to drift upward over time with real per capita income. Curious and curiouser…)

** update, 2/28 9pm Pacific:** A question has arisen as to whether WI and MN are very different. Here are depictions of sectoral value added for 2013 for the two states.

Can you tell the difference?

** Update, 3/1 noon Pacific:** I have calculated a similarity index for MN, IL, MI vs. WI, based on output composition. It’s an unweighted measure of absolute sector differences, (Σ|x

These four maps depict qualitative data applying to the 50 states plus the District of Columbia, to yield 51 observations. Blue means presence of the relevant institutional feature, gray means absence. I’ll consider four such features, call them Z1, Z2, Z3, and Z4.

**Figure 1:** United States, variable Z1=1 denoted by blue, variable Z1=0 by gray.

**Figure 2:** United States, variable Z2=1 denoted by blue, variable Z2=0 by gray.

**Figure 3:** United States, variable Z3=1 denoted by blue, variable Z3=0 by gray.

**Figure 4:** United States, variable Z4=1 denoted by blue, variable Z4=0 by gray.

Notice the interesting pattern. I convert the categorical data in these four maps into quantitative data using dummy variables, Z1 through Z4. Here are the correlation coefficients for the four variables, along with associated t-stats for the null hypothesis the correlation coefficient is zero.

Results reflect coding error fixed 2/25 5:20PM Pacific

Notice the correlation between Z2 and Z3 ~~are highest, at 0.61~~ is 0.55. The t-statistic for the null of zero correlation coefficient is soundly rejected at any conventional significance level. This confirms the impression gained by a visual inspection of maps 2 and 3; however, now we have a quantitative measure. (Note that the interpretation of a Pearson correlation coefficient when applied to binary variables is as a phi coefficient, also referred to as the “mean square contingency coefficient”.)

One can examine how the states align along each dimension by looking at a contingency table for Z2 and Z3 (notice the “phi coefficient” is the same as the correlation coefficient).

Results reflect coding error fixed 2/25 5:20PM Pacific

Eighteen states (35.3% of sample) fail to exhibit both of the characteristic in Z2 and Z3, while ~~23~~ 21 states (~~45.1%~~ 41.2% of sample) exhibit both of the characteristic in Z2 and Z3. A total ~~10~~ 12 states fall “off-diagonal”, with states exhibiting one, and not the other. That is, the correlation is not perfect.

One could estimate a linear regression between Z3 and Z2; this is called a linear probability model. Doing so would yield a slope coefficient of ~~0.62~~ 0.56, adjusted R-squared of ~~0.36~~ 0.29. This means a one unit increase in Z2 would increase the probability of Z3=1 from 0 to ~~0.62~~ 0.56. A linear probability model is problematic to the extent that it does not restrict probabilities to lie between 0 and 1. A probit model, which allows for a nonlinear relationship between Z3 and Z2 (and is based on the cumulative normal distribution) yields the following results:

Results reflect coding error fixed 2/25 5:20PM Pacific

The slope cannot be directly interpreted; one can find the implied probability using the cumulative normal distribution. When Z2 takes on a value of 0, then the probability of Z3=1 is 14.3%. When Z2 takes on a value of 1, then the probability is ~~76.7%~~ 70.3%.

Of course, even when one runs a regression, one can’t *necessarily* say one has identified a causal relationship, regardless of whether the coefficient is statistically significant or not. But certainly, knowing Z2 improves ones guesses of what Z3 will be. In fact, the above probit regression correctly predicts over ~~68.3%~~ 66.7% of the cases where Z3 takes on values of 0, and ~~88.5%~~ 87.5% of the cases where Z3 takes on values of 1 (assuming a cutoff value of 0.5, that is when the probability Z3=1 exceeds 0.5, predict Z3=1).

To highlight the non-causality interpretation, let’s consider what Z2 and Z3 are. Z3 takes on values of 1 if “right-to-work” laws are in effect, according to the National Right to Work Legal Defense Foundation, Inc.. Z2 takes on a value of 1 if anti-miscegenation laws were in effect in 1947. [1] A causal interpretation would be “having anti-miscegenation laws in 1947 cause one to have right-to-work laws in 2014”; this is clearly implausible. Reverse causality seems also implausible – that is it doesn’t seem likely that “having right-to-work laws in 2014 caused a state to have anti-miscegenation laws in 1947.” It is possible that adding in additional covariates would make the correlation disappear; but if it didn’t, a plausible interpretation is that there is a third, omitted, variable that caused certain states to have anti-miscegenation laws on the books in 1947, *and* caused certain states to have right-to-work laws in place in 2014.

By the way, the other variables are as follows: Z1 is a dummy variable that takes a value of 1 if restrictions on abortion at 20 weeks are in effect. [2] (some states have restrictions at 24 weeks, some third trimester, yet others at viability; and some had no restrictions.) I wanted to obtain a more general measure of restrictiveness on reproductive rights, but that would have entailed a lot more data collection, so I settled for this dummy variable. Finally, Z4 takes on a value of 1 if the state has implemented “stand-your-ground” laws. [3] Inclusion of these additional variables does not eliminate the statistically significant correlation found in the probit regression equation.

So…~~correlation is not causation!~~ to quote Edward Tufte, “Correlation is not causation but it sure is a hint.”

In spite of the current account reversals observed in advanced countries, global imbalances are still a matter of concern (IMF, 2014). Probably, the US current account is the most important component of such worldwide imbalances. The size of the US external deficit has been an issue of analysis for many years. Research on this specific topic has used, at least, two approaches. On the one hand, some researchers contend that thresholds in the dynamics of the current accounts exist. The simplest threshold model can be understood as one where a threshold value is used to identify ranges of values where the behavior predicted by the model varies in some relevant way. For example, Clarida, et al. (2005) find two thresholds in the US current-account-to-GDP ratio. According to that, if the current account surplus is above 2.2% or below -2.2% of GDP, we should expect a reversal toward its long-run mean. Usually, these papers employ only the information contained in the time series of the current account itself (a univariate approach). On the other hand, a number of works based on dynamic stochastic general equilibrium (DSGE) models suggest that the US current account is driven by shocks of fiscal balance, productivity level, productivity volatility, or oil prices.^{1}

Even though a nonlinear univariate model might be useful –for example, for forecasting purposes– its nature leaves aside the fundamentals behind the current account dynamics. The DSGE literature, in turn, usually focuses on linear(ized) relationships and emphasizes one or two factors –partly due to the curse of dimensionality. In the paper titled “A Threshold Model of the US Current Account”, I aim to bridge the gap between these two branches in a multivariate nonlinear framework that can offer more tractability. I address several questions: What are the main drivers of the US current account? Is the behavior of the current account the same during deficits and surpluses or does the size of the external imbalance matter as some analysts suggest? Is there a threshold relationship between the current account and its drivers?

To answer these questions, I estimate a threshold model with multiple regressors to explain the behavior of the US current account during the period between 1973.I and 2012.I, and test for the presence of regimes in its dynamics. As threshold candidates, I try a set of variables suggested by commentators and previous empirical works: the (lagged) level and the size of the current account-to-GDP ratio, the (lagged) level and the size of the fiscal-balance-to-GDP ratio, and time. In the latter case, I actually test for the presence of an unknown time break in the relationship between the regressors and the dependent variable. As regressors, I evaluate a similar set to the one proposed in the DSGE literature mentioned above. In addition, I include the real interest rate and the real exchange rate.^{2}

To deal with the potential endogeneity of the regressors, I use the IV estimator of a threshold model developed by Caner and Hansen (2004).

*Main findings*

First, in contrast to the univariate threshold models, time is the most important threshold variable. I find a robust time break –not previously documented in the literature– in the relationship between the current account and its main drivers in the third quarter of 1997. Our estimates stubbornly point to 1997.III as the time break even if I use a larger sample such as 1957.I-2012.I.

The time break found in 1997.III coincides with two events: the onset of the Asian financial crisis and the Taxpayer Relief Act of 1997. The former implied a recomposition of portfolios among international investors, including central banks, and the decision of sharp devaluations, the imposition of capital controls, and reserves buildup by monetary authorities. In addition, the Asian financial crisis is viewed as the onset of a sequence of international crises among emerging market economies. Other economies that faced similar crises were Russia (1998), Brazil (1998), Argentina (1999-2002), and Turkey (2001). All of them involved sharp devaluations, modification of the exchange rate regime, and the rise of foreign exchange reserves as a hedge against potential speculative attacks or another financial crisis. While the change in exchange rate policies to limit currency appreciations has led to some to talk about a revived Bretton Woods system (Dooley et al., 2003), the *war chests* of foreign reserves have, at least in part, led to an increasing purchase of US treasury bonds, which is usually linked to the so- called *global saving glut hypothesis* (Bernanke, 2005). The second factor that might have contributed to the structural change originated domestically. The Taxpayer Relief Act of 1997, enacted on August 5th, reduced several federal taxes, provided some tax exemptions, and extended tax credits. According to estimates posted by the NBER, the average marginal tax on long-term gains was reduced by almost 7%, from 25.6% to 18.7% in 1997, the largest cut since 1960.

Second, as opposed to what other authors contend, I did not find evidence on the importance of the size or the sign of the current account as threshold variables. The time line always dominates any potential threshold variable previously used or proposed by the empirical literature in terms of model fit. The other candidate variables not only fail to provide an adequate fit compared to the time line, but also they are highly sensitive to the sample, and did not provide either precise threshold or coefficient estimates, or statistics that could support a valid model in each regime.

Third, the most significant determinants of the US current account are total factor productivity, the real exchange rate, the fiscal surplus, and the volatility of productivity in both regimes (before and after 1997.III). As the paper shows, the most statistically significant shifts are related to the productivity level, the real exchange rate, and the real interest rate. In particular, productivity shocks became more important after 1997.

The figure above displays the economic significance of each regressor. That is, the coefficient estimate multiplied by the standard deviation of the corresponding regressor. For example, one-standard-deviation shock in productivity lowers the current account in approximately 0.15 points of long-run GDP in the pre-1997 regime, whereas the respective reduction is 0.3 points of long-run GDP in the post-1997 regime.

Fourth, the relative price of oil and the real interest rate become statistically significant and more economically relevant after 1997, although shocks to these relative prices do not contribute significantly to fluctuations in the current account surplus, at least as much as other regressors of the model. For example, a one-standard-deviation shock in the interest rate would be associated with a rise in the current account of 0.11% of long-run GDP. From an economic viewpoint, the saving glut effect through the interest rate is relatively less important. Similarly, a one-standard-deviation shock in oil prices above its trend would be related to a current account decline of 0.11 percentage points of long-run GDP.

*Why did productivity shocks become more important after 1997?*

We mentioned that the time break coincides with the onset of the Asian financial crisis. One possibility is that international investors moved their funds from East Asia and, perhaps, other emerging market economies, to the US and invested in more capital-intensive sectors such as the information industry.^{3} The best example of this investment shift could have been the dot-com bubble observed between 1997 and 2000. The capital intensification of the economy could have made productivity shocks a more important driver of investment and, as a result, the current account. To a lesser degree, another possibility is that the Taxpayer Relief Act of 1997 raised the sensitivity of domestic absorption and, consequently, the sensitivity of the current account to productivity shocks.^{4}

To summarize, we find that time is the best threshold variable in the dynamics of the US current account. In particular, one regime exists before and another one exists after the third quarter of 1997, a period that coincides with the onset of the Asian financial crisis and the Taxpayer Relief Act of 1997. Productivity has become a more important driver of the US current account since then. Further research is needed to verify the reason behind this fact. An important implication for practitioners, who seek to improve the fit of their models that attempt to explain the current account deficit, is the need of taking into account the 1997 structural break and modeling it in a DSGE framework. Such task could be one of the next steps in the research agenda on global imbalances.

*References*

Acemoglu, D., Guerrieri, V., 2006. Capital deepening and non-balanced economic growth. NBER Working Paper 12475.

Bernanke, B., 2005. The Global Saving Glut and the U.S. Current Account Deficit. Remarks at the Sandridge Lecture, Virginia Association of Economists, Richmond, Virginia. The Federal Reserve Board.

Bodenstein, M., Erceg, C.J., Guerrieri, L., 2011. Oil shocks and external adjustment. J. Int Econ. 83, 168-184.

Bussiere, M., Fratzscher, M., Muller, G., 2010. Productivity shocks, budget deficits and the current account. J. Int. Money Financ. 29, 1562-1579.

Caner, M., Hansen, B., 2004. Instrumental Variable Estimator of a Threshold Model. Economet. Theor. 20, 813-43.

Chinn, M., Prasad, E., 2003. Medium Term Determinants of Current Accounts in Industrial and Developing Countries: An Empirical Exploration. J. Int Econ. 59(1), 47-76.

Clarida, R., Goretti, M., Taylor, M., 2005. Are There Thresholds of Current Account Adjustments? NBER conference G7 Current Account Imbalances: Sustainability and Adjustment.

Dooley, M., Folkerts-Landau, D., Garber, P., 2003. An Essay on the Revived Bretton Woods System, NBER Working Paper No. 9971.

Duncan, R., 2015. A Threshold Model of the US Current Account, forthcoming in Economic Modelling.

Fogli, A., Perri, F., 2006. The Great Moderation and the U.S. External Imbalance. Monetary and Economic Studies (Special Edition). 209-234.

International Monetary Fund, 2014. World Economic Outlook, Legacies, Clouds, Uncertainties (October).

^{1. See Bussiere, et al. (2010), Fogli and Perri (2006), and Bodenstein, et al. (2011), respectively. Another branch of the empirical literature centers its attention to medium-term fluctuations of the current account using cross-country samples (e.g., Chinn and Prasad, 2003) and overlaps with the DGSE branch. The inclusion of demographic regressors, for example, is more appropriate in cross-country regressions rather than time-series models due to their low variability over time. }

^{2. A reduction in this index indicates real currency depreciation. }

^{3. According to Acemoglu and Guerreri (2006), the information sector in the US has a capital share of 0.53 (the average capital intensity is around 0.4). }

^{4. Consider, for simplicity, an economy in which a productivity shock raises dividends and, therefore, generates capital gains. If capital gains are taxed at the rate t, then consumption would increase by a proportion that depends on 1-t. If such tax rate is reduced, the sensitivity of consumption to productivity shocks would increase. }

Here’s the program for CBO at 40

]]>Welcome

Douglas W. Elmendorf, Director

Opening Remarks

Representatives of the Budget Committees

Keynote

Alice Rivlin, Founding Director

Panel Discussion

A panel of former Congressional Budget Office directors will discuss CBO’s past and future. The panelists will also respond to questions from the audience.

Alice M. Rivlin, Director 1975-1983,

Senior Fellow, Brookings InstitutionRudolph G. Penner, Director 1983-1987,

Senior Fellow, The Urban InstituteRobert D. Reischauer, Director 1989-1995,

President Emeritus, The Urban InstituteJune E. O’Neill, Director 1995-1999,

Professor of Economics, Baruch CollegeDan L. Crippen, Director 1999-2003,

Director, National Governors AssociationDouglas Holtz-Eakin, Director 2003-2005,

President, American Action ForumPeter R. Orszag, Director 2007-2008,

Vice Chairman of Corporate and Investment Banking, CitigroupClosing Remarks

Jim Guest says the bill would serve Americans’ “right to know where their tax dollars are going.” Perhaps he meant to say Americans’ right to know where the Treasury’s revenues are coming from rather than where tax dollars are going. The Federal Reserve’s net contributions to the U.S. Treasury have averaged +$83 billion per year since 2009. Last year’s federal deficit would have been almost $100 billion bigger if it had not been for the net positive revenue contributions from the Fed.

John Tate thinks the bill would help address “the silent, destructive tax of monetary inflation.” But inflation as measured by the consumer price index has averaged under 1.8% over the last decade. That’s the lowest it’s been since the 1960s.

Of course, many of the same people who favor Senator Paul’s bill distrust government-collected inflation data like the CPI. So suppose you look at the private Billion Prices Project, which mechanically collects a huge number of prices each day off the internet. According to BPP, inflation over the last year has been if anything lower than the official numbers.

Others may take the view that more transparency in and of itself is a good thing. But the Fed is already audited; you can read the audit yourself here. You can examine the Fed’s assets directly down to the level of CUSIP, if you like. Here at Econbrowser we’ve been reporting detailed graphs of the Fed’s assets and liabilities for years using publicly available sources like the weekly H41 statistical release. Cecchetti and Schoenholtz note this conclusion from Richard Fisher, president of the Federal Reserve Bank of Dallas and one of the FOMC’s outspoken critics of quantitative easing:

We are– I’ll be blunt– audited out the wazoo. Every Federal Reserve Bank has a private auditor. We have our auditor of the system. We have our own inspector general. We are audited. What he’s talking about is politicizing monetary policy.

The Wall Street Journal’s David Wessel elaborates:

In 2009, Congress changed the law to allow GAO audits of loans made by the Fed to a single company, such as Bear Stearns or Citigroup, but only when the Fed invoked Section 13(3) of the Federal Reserve Act. (That’s the provision that allows the Fed to lend to almost anybody under circumstances it deems “unusual and exigent.”) The Dodd-Frank law of 2010 further widened the GAO’s authority, allowing it to review the Fed’s internal controls, policies on collateral, use of contractors and other activities—but the GAO is still blocked from reviewing or evaluating the Fed’s monetary-policy decisions.

The Audit the Fed bill would change the law again, and allow the GAO to examine and criticize all monetary policy decisions without restriction.

If we did want to see a lot more inflation for the U.S., Paul’s bill would be the way to get it. The main effect of the bill would be to give Congress an additional tool to exert operational control over monetary policy. The political pressures will be very strong not to raise interest rates when the time does come to start to worry again about inflation. And when the Fed does get to raising rates, it will mean extra costs for the Treasury in paying interest on the federal debt– Congress isn’t going to like that. The primary effect of the legislation would be to give Congress one more stick with which to try to beat up on the Fed when the Fed next does need to take steps to keep inflation from rising.

In fact there’s a pretty dependable historical correlation– the more political control over monetary policy that a country gives to the legislature and the administration, the higher the inflation rate the country is likely to get.

Senator Paul’s bill is unambiguously a bad idea.

]]>

*Larry Kudlow*

From The Financial Crisis: Foreseeable and Preventable (Feb. 2011). Jeff Frieden asks, in the *NY Times*, why warnings of imminent housing collapse and financial crisis were ignored:

Ideology probably mattered. Larry Kudlow, economics editor of the conservative National Review, in 2005 dismissed “all the bubbleheads who expect housing-price crashes in Las Vegas or Naples, Florida, to bring down the consumer, the rest of the economy, and the entire stock market.” Of course, the bubbleheads were exactly right, but the predictions did not accord with Kudlow’s partisan commitments or his ideology.

And so it is with the post-mortems. Politicians, special interests, and ideologues all have their reasons to insist on a particular interpretation of the crisis. And those connected to the Bush administration have strong incentives to deny that the administration could have done anything differently. But they are wrong.

*Stephen Moore*

From State Employment Trends: Does a Low Tax/Right-to-Work/Low Minimum Wage Regime Correlate to Growth?, it’s shown that the Laffer-Moore-Williams *Rich States, Poor States* ranking of business environment does *not* correlate with growth. 47th ranked California outpaces 17th ranked Wisconsin (or 15th ranked Kansas). Using the entire 50 state ranking, I also show that there is little apparent correlation.

**Figure 1:** Ranking by annualized growth rate in log coincident index 2013M01-2014M03 versus 2013 ALEC-Laffer “Economic Outlook” ranking. Nearest neighbor nonparametric smoother line in red (window = 0.7). Source: Philadelphia Fed, ALEC, and author’s calculations.

*Arthur Laffer*

I first met Arthur Laffer more than 30 years ago. His presence on Econbrowser has been consistent, most recently in Whistling past the intellectual graveyard…the Extreme Supply-Sider one in Topeka, that is. Additional appearances, on seasonals (joint appearance with Professor Casey Mulligan), and in spirit, supply side responses (joint appearance with Bill Beach/Heritage Foundation), tax elasticities (joint appearance with Governor Mitt Romney).

I am ever thankful for the likes of Kudlow, Moore, and Laffer, even as they drag down the level of economic discourse. They just provide too many examples of how not to conduct serious analysis.

(For those who don’t recognize the allusion in the title, see here.)

*Update, 8PM Pacific:*

NB: **Rick Stryker** Notes that these men are not formally Governor Walker’s economic advisers. That observation is correct; they merely hosted the private meeting that Walker was guest of honor and provided their advice, as discussed here. The Governor has a set of official economic advisers in Wisconsin state government.

** Update, 2/22, 9:30PM Pacific:** Here is an article on one of the Governor’s economic advisers.

** Update, 2/25, 2:30PM Pacific:** Here is a Bloomberg article on the Wisconsin economy’s progress under Governor Walker.