Category Archives: Econ Data

GDP Growth is Strong – Perhaps Too Strong

Posted on by

A.  Introduction

On April 25, the Bureau of Economic Analysis released its initial estimates of the GDP accounts for the first quarter of 2024 – what it calls the “Advance Estimate”.  These initial estimates of the growth of GDP and of its components are eagerly awaited by analysts.  While revised and updated in subsequent months as more complete data become available, it provides the first good indication of what recent growth has been.

In the first quarter of 2024, real GDP grew at an estimated annual rate of 1.6%.  This was viewed by many analysts as disappointing, as the average expectation (based on a survey of economists by Dow Jones) was for 2.4% growth.  And it was a deceleration in the rate of growth of GDP from 4.9% in the third quarter of 2023 and 3.4% in the fourth quarter.  The Dow Jones Industrial Average then fell by 720 points in the first half hour of trading (1.9%), with this attributed to the “disappointing” report on GDP growth.  It later recovered about half of this during the day.

One should have some sympathy for the commentators who are called upon by the media to provide an almost instantaneous analysis of what such economic releases imply.  But if they had examined the release more closely, the conclusion should not have been that economic growth was disappointingly slow, but rather that it has been sustained at a surprisingly high level.  After two quarters of extremely fast growth in the second half of 2023, some moderation in the pace should not only have been expected but welcomed.

The economy under Biden has been remarkably strong.  The unemployment rate has been at 4.0% or less for 28 straight months, and has reached as low as 3.4%.  Unemployment has not been this low nor for this long since the 1960s.  And an economy at full employment can only grow at a potential rate dictated by labor force growth and productivity.  The ceiling is not a hard one in any one quarter (labor utilization rates, and hence productivity, can vary in the short term), plus there is statistical noise in the GDP estimates themselves.  But growth in what is called “potential GDP” sets a ceiling on what trend growth might be.  And if the economy is at or close to that ceiling (as it is now), it can only grow over time at the pace that the ceiling itself grows at.

B.  Potential GDP

There are various ways to determine what potential GDP might be.  A respected and widely cited estimate is produced by the Congressional Budget Office (CBO), with figures on potential GDP for both past and future periods (up to 10 years out).  It is based on estimates of what the potential labor force has been or will be, accumulated capital, and technological progress.  In any given year, the CBO estimates reflect what GDP could be with the capital stock that would be available and the production that capital would allow, along with labor utilization at “full employment”.

The chart at the top of this post shows what real GDP per capita has been over the 11 years from the start of Obama’s second term (2013Q1) to now (2024Q1), along with the estimate by the CBO of potential GDP (expressed in per capita terms).  Not only is the economy now close to the potential GDP ceiling, it is a bit beyond it.  This is possible with the CBO estimates of potential GDP as they assume the labor market cannot sustain for long an unemployment rate of below roughly about 4.5% (which can vary some over time, based on the structure of the labor force).  Hence if actual unemployment is below this – as it is now and as it was for a period in 2019 – “potential GDP” as estimated by the CBO can be below actual GDP.  There are other factors as well, but the level of unemployment is the most significant.

This is also why actual GDP was below potential GDP from 2013 to late 2017 in the chart.  Unemployment was still relatively high in 2013 coming out of the 2008/09 economic and financial collapse.  As discussed in earlier posts on this blog (see here and here) limitations on government spending imposed by the Republican-controlled Congress slowed the recovery from that downturn and kept GDP well below potential for far too long.  This was the first time government spending had been cut following a recession since the early 1970s.  Federal government spending on goods and services fell at an average annual rate of 3.2% each year (in real terms) from 2011 to 2014.  In 2015 and 2016 it was finally allowed to grow, but only at a slow 0.3% per year pace on average.  Only after Trump was elected did Congress allow federal government spending to rise at a more significant rate – at 2.6% per year between 2017 and 2019 (and then by much more in 2020 due to the Covid crisis, by which time Democrats controlled Congress).

This lack of a supportive fiscal policy following the 2008/09 economic and financial collapse slowed the pace of recovery.  Unemployment fell only slowly, but did still fall, and reached 4.7% by the end of Obama’s second term.  The gap between actual and potential GDP diminished, and as seen in the chart at the top of this post, actual GDP has been close to potential GDP since 2018 (with the important exception of the 2020 collapse due to Covid).

The economy is now at – or indeed a bit above – the CBO estimate of potential GDP.  Although there may be quarter-to-quarter fluctuations – as noted above – going forward one cannot expect GDP to grow on a sustained basis faster than that ceiling.  And the CBO forecasts that potential GDP is growing at a 2.2% pace currently, with this expected to diminish over time to a 2.0% pace by 2030 and a 1.8% pace by 2034.  This is primarily due to demographics:  Growth in the labor force is slowing.

Real GDP grew at an average annual rate of 4.1% in the second half of 2023 (3.4% in the third quarter and 4.9% in the fourth quarter).  This is well above the CBO’s estimate of potential GDP growing at a 2.2% rate.  Some slowdown should have been expected.  Even with the 1.6% rate for the first quarter of 2024, real GDP has grown at an average annual rate of 3.3% since mid-2023.  It should not be surprising if GDP growth in the second quarter of 2024 comes in at a relatively modest rate, as the economy returns to the trend growth that potential GDP allows.

However, the initial indication from the Atlanta Fed’s GDPNow indicator is that GDP growth in the second quarter of 2024 will in fact be quite high at a 3.9% rate (in its initial estimate made on April 26 – the most recent as I write this).  If that turns out to be the case, it would not be surprising if the Fed becomes concerned with a pace of growth that is excessively fast.

C.  Other Indicators of an Economy Fully Utilizing Its Potential

One wants an economy that is fully utilizing its potential.  With full employment, one is not throwing away goods and services – as well as the corresponding wages and income – that labor and producers would be eager and able to provide.  But once an economy has reached that potential, it can only grow over time at the rate that that potential grows.  This pace is dictated by demographics (growth in the labor force) and growth in productivity.  While this will be a slower pace than what would be possible for an economy with underutilized labor and other resources – where a period of more rapid growth is possible by bringing into employment those underutilized resources – once one is at the ceiling one cannot grow on a sustained basis at a pace higher than that.  Trying to do so leads to inflation.

With this perspective, a number of observations come together from this release on 2024 first-quarter growth in GDP and its components:

a)  The 1.6% growth rate was viewed as “low”.  But as noted above, this followed exceptionally high growth, of 4.9% in the third quarter of 2023 and 3.4% in the fourth quarter.  One should have expected a slowdown.

b)  There is also evidence of the economy reaching its capacity limits in how the particular components of the GDP figures changed.  Keep in mind that GDP, while derived in these accounts from estimates of what was sold for final demand uses (consumption, investment, etc.), is still a measure of production, not just sales.  That is, GDP – Gross Domestic Product – is a measure of what is produced, produced domestically, and in “gross” terms (because investment is counted in gross terms rather than net of depreciation).

The reason this indirect approach to estimating production works is because whatever is produced and not sold will end up as an increase in inventories.  And this change in inventories is treated as if it were a final demand category.  It can be viewed as a form of investment (investment in inventories), and is included in the accounts as part of overall investment (i.e. it is added to fixed investment, which is investment in machinery and structures).

Furthermore, foreign trade is included in net terms:  exports less imports.  Part of what is produced domestically is sold for exports, while imports supply products that can be used to satisfy domestic demands.

When an economy is operating at or close to potential GDP, one can expect final demands to be increasingly met by drawdowns of inventory (or less of an increase in inventories compared to before) plus a decline in the net trade balance (less exports and/or more imports).  Each can supply product to meet final demands when domestic production is constrained because the economy is operating at close to the ceiling.

One sees both of these in the 2024Q1 figures.  While inventories still rose (by $35 billion, in 2017 constant prices), they rose by less than they had in 2023Q4 (when they rose by $55 billion).  Thus, while GDP includes the change in inventories as one of the demand components along with consumption and other investments, the change in GDP will be based on the change in the change in inventories.  (See this earlier post on this blog.)  And that fell in 2024Q1, as inventory accumulation – while still positive – was not as high as it had been in the previous quarter.  That change in the inventories component reduced GDP growth by 0.35% points relative to what it would have been had domestic production been such that inventory accumulation would have matched what it had been in the preceding quarter.  That it did not can be a sign that domestic production is being constrained by capacity.

Similarly and more importantly, the net trade balance fell.  While exports grew slightly (0.1% of GDP) imports rose by much more (1.0% of GDP), and hence the net trade balance fell by 0.9% of GDP.  This is consistent with domestic production being constrained by an economy that was already at full employment and could not immediately produce much more, and hence with demand that was increasingly met by net imports.

The changes in the net trade balance and in net inventory accumulation totaled 1.2% of GDP (before rounding).  That is, production (GDP) would have had to increase by 2.8% rather than 1.6% to supply domestic purchasers of final (i.e. non-inventory) product.  But with production constrained by capacity limits, the economy had to import more and limit inventory accumulation to less than before.

I should emphasize that this is not a bad position to be in.  One wants an economy operating at full capacity.  But when the economy is operating at full capacity, there will be limits on how much can be supplied domestically.  And as noted before, one cannot expect growth going forward – on average and recognizing there will be period-to-period fluctuations – to exceed the rate at which potential GDP can grow.

c)  Another indication of an economy reaching its potential ceiling is what is happening to prices.  This is more disconcerting.  Price deflators are estimated as part of the GDP accounts in order to convert (deflate) the nominal estimates of the various GDP components into estimates of what the real changes were.  While people focus on changes in real GDP and its components – and properly so – some may not fully realize that the data the BEA collects on production and sales are all in nominal money terms.  It is not really possible for producers to report anything else.  The BEA then converts those nominal money figures to changes in real terms by applying price indices to “deflate” the nominal figures – hence the term “deflator”.  The BEA obtains those price indices – tens of thousands of them – separately, primarily from the price surveys carried out by the Bureau of Labor Statistics.

The initial estimates of the GDP accounts released on April 25 indicated that the price deflators for both overall GDP and for the Personal Consumption Expenditures (PCE) component of GDP demand rose at higher rates than in the preceding several quarters.  The GDP deflator rose in the first quarter of 2024 at an estimated annual rate of 3.1% and the PCE deflator at a rate of 3.4%.  The PCE deflator receives special attention as it is the primary measure of inflation that the Fed focuses on as it considers what monetary policy to follow.  The Fed pays attention to much more as well, of course, but the PCE deflator is special.  And the Fed target for the PCE deflator is 2.0%.

The annualized rates for the GDP and PCE deflators were at 1.6% and 1.8%, respectively, in the fourth quarter of 2023.  They had been generally coming down since mid-2022, and had averaged 2.2% and 2.3% respectively in the final three quarters of 2023.  The increase in the first quarter of 2024 was therefore of some concern, especially when coupled with the other indications (discussed above) that the economy is now at or even above the potential GDP ceiling.

But it is also important to keep in mind that – as often said – one period’s figures do not constitute a trend.  There have been, and will be, quarter to quarter fluctuations.  But the increase in the price deflators from below the Fed’s 2.0% target to a level a good deal higher, coupled with the other indications of an economy operating at or close to capacity, is something to watch.  And it suggests that the Fed is likely to remain cautious and not reduce interest rates from where they now are until they find out more about what is happening to prices.

D.  The Federal Fiscal Deficit is Large

Finally, while not part of the report on the GDP accounts, it should be noted that the federal fiscal deficit remains extremely high.  Recent figures on the Federal Government’s fiscal outlays, receipts, and deficit, expressed here as a share of GDP in the periods, are as follows:

Federal Government Fiscal Accounts

GDP shares

Receipts

Outlays

Deficit
FY2023

16.5%

22.7%

6.3%
CY2023

16.5%

23.0%

6.5%
FY2023 H1

15.4%

23.7%

8.3%
FY2023 H2

17.5%

21.8%

4.3%
FY2024 H1

15.6%

23.1%

7.6%

The GDP shares are calculated from the dollar figures reported in the Monthly US Treasury Statement for March 2024, coupled with the GDP estimates of the BEA.  The Monthly Treasury Statements are definitive in that the reported dollar figures up to the current month rarely change later (although forecasts for the full budget year of course may).  Note also that the reported monthly figures are not seasonally adjusted but are rather the actual fiscal receipts and outlays for the period, while the GDP figures are seasonally adjusted.

In a period of full employment, these deficit figures are all high.  As was discussed in an earlier post on this blog, while high fiscal deficits may well be necessary and appropriate when unemployment is high, one should balance this with lower deficits when the economy is at full employment – as it is now.  The fiscal deficits need not be zero, but a good rule of thumb is to aim for a deficit of perhaps 3% of GDP and no more than 4% of GDP in an economy that is at full employment.  At such deficits, the government debt to GDP ratio will be stable or falling over time, which can then balance out the times when the appropriate policy is to allow for a higher deficit in an economic downturn in order to support a recovery.

The math is simple.  As of March 31, 2024, the total federal debt held by the public was $27.5 trillion (as reported in the Monthly Treasury Statement).  Nominal GDP in 2024Q1 was $28.3 trillion (at an annual rate).  The debt to GDP ratio was thus 97.3% (before rounding), or close to 100%.  If, going forward, one should expect trend growth of about 2% per year in real GDP, inflation of 2% (the Fed’s goal), long-term Treasury interest rates of 4% (i.e. 2% inflation and a 2% real rate of interest on longer-term securities), then a debt to GDP ratio of 100% will stay at 100% if the federal fiscal deficit is 4% of GDP.  The debt ratio will fall with a lower deficit and rise with a higher deficit.

But despite being at full employment, the federal fiscal deficit was 7.6% of GDP in the first half of FY2024.  That is well above the 4% level needed to keep the debt to GDP ratio from rising further.  However, It is not clear whether the deficit has been trending higher or lower.  While the 7.6% deficit in the first half of FY2024 was higher than the 6.3% deficit in FY2023 as a whole, and substantially higher than the 4.3% deficit in the second half of FY2023, it is less than the 8.3% deficit in the first half of FY2023.  There is likely a significant degree of seasonality in the fiscal figures.  But under any reasonable scenario, the deficit will be well above 4% of GDP again this fiscal year.

The issue facing the Democrats is that every time over the past more than 40 years that they have cut the fiscal deficit during their term in office, the subsequent Republican administration has then increased it – through a combination of tax cuts and expenditure increases.  Comparing fiscal years (and avoiding recession years given their special nature, and based on data from the CBO), the fiscal deficit under Ford in FY1976 was 4.1% of GDP.  Carter brought that down by FY1979 to just 1.6% of GDP.  Reagan tax cuts and expenditure increases then raised the deficit to 5.9% of GDP in FY1983, and it was 4.5% of GDP under Bush I in FY1992.  The fiscal accounts then moved into a surplus under Clinton following the steady and strong growth in real GDP during his presidency, reaching a surplus of 2.3% of GDP in FY2000.  On taking office, Bush II at first advocated tax cuts because the economy was strong and the fiscal accounts were in surplus, but then after the downturn a few months after taking office, Bush II promoted tax cuts because the economy was weak.  The tax cuts did go through, and with fiscal revenues falling as a share of GDP while expenditures rose, the fiscal deficit reached 3.4% of GDP in FY2004 – a huge shift of 5.7% points of GDP from where it was in Clinton’s last year in office.

With the economic and financial collapse in 2008 in the last year of the Bush II presidency, the deficit rose to 9.8% of GDP in FY2009 in Obama’s first year.  This stabilized an economy that had been in freefall as Obama took office (with the sharpest downturn since the Great Depression), but as noted above, subsequent cuts in government spending then slowed the full recovery.  Eventually the economy did recover, and the fiscal deficit was reduced to 2.4% of GDP in FY2015 and a somewhat higher 3.1% of GDP in FY2016 when federal government spending was finally allowed to grow, albeit modestly.

Taxes were then once again cut under the Republican presidency of Trump, and despite an economy at full employment, the fiscal deficit rose to 4.6% of GDP in FY2019.  It then exploded with the Covid crisis, to 14.7% of GDP in FY2020 and 12.1% in FY2021, before falling under Biden to 5.4% of GDP in FY2022 and 6.3% of GDP in FY2023.

So what should be done?  This is not the place for a full analysis, but broadly, fiscal revenues as a share of GDP are low in the US.  Total tax revenue (including by state and local governments) is lower in the US than in any other high-income member of the OECD with just one exception (Switzerland), with US tax revenues more than 6% points of GDP less than the OECD average (in 2022).  A post on this blog from 2013 – now perhaps out of date – showed that the federal government debt to GDP ratio would have fallen sharply – rather than increase – in the years then following if the Bush II tax cuts had been allowed to expire in full at the end of 2012.  The figures would be different now, but the basic point remains that both compared to other high-income nations and to the historical record, the US suffers from a chronic fiscal revenue problem.

A reasonable target for federal fiscal revenues might be 20% of GDP – the same share of GDP as in FY2000.  That would be an increase of 3.5% of GDP from the 16.5% collected in FY2023.  Taxes collected in the US would still be less – as a share of GDP – of all but two of the higher-income OECD members (Australia and Switzerland), and also far less than the OECD average.

There are also always some fiscal expenditures that could also rationally be cut (but where there is always disagreement on which), but even with no cuts in expenditures, revenues of 20% of GDP in FY2023 would have brought the deficit down from 6.3% of GDP to 2.8%.  And as discussed above, a deficit of 2.8% of GDP would be expected to lead to a downward trend over time in the government debt to GDP ratio.

One option to get fiscal revenues back to around 20% of GDP would be simply to bring back the taxation rules of that year.  They were not excessively burdensome – the economy was performing well at the time with solid GDP growth and low unemployment.  But better would be to introduce true tax reforms, such as ending the disparities in the tax system where different forms of income are taxed differently (as discussed, for example, in this earlier post on this blog).  The most significant such disparity is that income from wealth (which is, not surprisingly, mostly held by the wealthy) is taxed at lower rates than income from wages.  But with Republicans in control of Congress, such a reform would never be passed.

E.  Summary and Conclusion

The economy is at full employment and is producing at or close to the ceiling allowed by its productive potential.  Going forward, one should not expect growth in real GDP to be greater than the pace at which this ceiling grows.  There may well be quarter-to-quarter fluctuations around this, as the ceiling is not absolute (labor utilization can vary) plus there is statistical noise in the GDP estimates themselves, but over time one should expect – and indeed welcome – growth that averages what that ceiling grows at.  The CBO estimates that potential GDP is growing at a rate of about 2.2% per annum currently, and expects this to fall over time to a 2.0% rate by 2030.

The 1.6% rate of growth in the first quarter of 2024 should be seen in this light.  Real GDP had grown at rates of 4.9% in the third quarter of 2023 and 3.4% in the fourth quarter, and a slowdown from such a pace should not only have been expected but welcomed.

Indeed, there may be a concern that GDP growth has been too rapid since mid-2023.  Even with the 1.6% growth of the first quarter of 2024, growth has averaged 3.3% since the middle of last year.  And there are signs in the GDP accounts themselves of an economy producing at capacity.  Inventory accumulation slowed relative to what it was before while the foreign trade balance fell as imports rose substantially.  The deflators for GDP and for Personal Consumption Expenditures also rose – to annualized rates of 3.1% and 3.4% respectively – after following a downward trend since mid-2022.  This is, however, an increase for the deflators for just one period at this point, and one should not assume until there is further evidence whether this marks a change in that previous trend.

For an economy at full employment, the current size of the fiscal deficit is a concern.  At full employment one should be aiming for a deficit of below around 4% of GDP in order at least to stabilize and preferably reduce the government debt to GDP ratio.  But in FY2023, the deficit was 6.3% of GDP.  The US has been facing chronic deficit issues for decades now – a consequence of the tax cut measures pushed through by Reagan, Bush II, and Trump.  A reasonable goal now would be a tax reform that removes the distortions from taxing different types of income differently, with rates then set to obtain fiscal revenues of around 20% of GDP – an increase of 3.5% points of GDP compared to the revenues collected in 2023.  The tax rates on income from wealth would rise from the preferential rates they now enjoy, while the tax rates on income from wages (and other “ordinary income”) might well fall.

Even with such an increase, fiscal revenues collected would still be well below the OECD average, and below that of all but only two of the higher-income OECD members.  In contrast, cuts in expenditures (as was done, as a share of GDP, during the presidencies of Carter, Clinton, and Obama), are likely to be followed in the next Republican administration with another round of tax cuts.

Inflation in the US Would Meet the Fed Target of 2% if Calculated as Europe Does

Posted on by

No price index is perfect.  Assumptions need to be made on what to include and how to include it.  Based on those decisions, the resulting price indices (and hence inflation rates) can differ and differ significantly.  And this can affect policy.

In this context, it is interesting to compare what inflation would be when calculated as the US does for the widely followed consumer price index (CPI), or if it were calculated according to the standard followed in the European Union for what it calls the harmonized index of consumer prices (HICP).  Both are reasonable measures, but the resulting inflation can be quite different, as seen in the chart above.  With the CPI, the Fed may conclude inflation is still too high – above its 2% target.  But calculated as Europe does, one could conclude that inflation is now too low.

This short post will look at the differences and the primary reasons for them.  There are lessons to be learned.  In particular, it is important to understand what lies behind various statistical measures – including, but not only, any measure of inflation – and not blindly focus on just one when arriving at policy decisions.  The Fed in general does, and the Fed’s Board has an excellent staff to advise on developments in the economy.  But the media often does not consider such distinctions.

The chart at the top of this post shows the 6-month rolling average percentage changes in prices (at annualized rates) for the period from December 2020 through to January 2024.  Both measures are for the US, and both are calculated by the Bureau of Labor Statistics (BLS) based on the same data on prices that the BLS collects.  The CPI data can be found here, while US inflation based on the HICP methodology as calculated by the BLS can be found here.  The BLS notes that its calculations of US inflation based on the HICP methodology are carried out outside of the “official production system” (as it calls it), and are more in the nature of a research project.  But the BLS uses the same underlying data for the HICP measure as it uses for its regular CPI calculations.

The HICP methodology was developed as Europe moved to greater monetary integration, culminating in the creation of a common currency – the euro – as well as the European Central Bank (the ECB).  The ECB has – similarly to the Fed – the objective of targeting a 2% rate of inflation.  For this, it obviously needs to know what inflation is in the Eurozone.  But the member nations of Europe that came together to adopt the euro as a common currency (currently 20 nations) had each long had their own way of estimating inflation within their countries, with various methodologies used.

A common approach needed to be adopted, and starting with regulations issued in 1995, the participating nations agreed to what was labeled the “harmonized index (or indices) of consumer prices” (HICP).  The statistical agencies of the EU member countries would follow that common methodology, and report their results to Eurostat for aggregation across the countries to a euro-wide index of inflation for use by the ECB.  The HICP is now used also for international comparisons of inflation, and it is in this context that the BLS prepares its HICP inflation index for the US.

There are a number of differences between the approaches used for the HICP and for the CPI that lead to the differences in the inflation rates seen in the chart above.  The key ones are:

a)  The HICP only includes prices of goods and services where there are direct monetary expenditures.  The CPI, in contrast, includes estimates of what the implicit costs are of certain services where there are not such direct expenditures.  The most important of these are the services provided in owner-occupied homes.  The CPI assumes that rents are implicitly being paid at rates similar to what is being paid by those who actually do rent.  As was discussed in a post on this blog from last May, the way rents are adjusted (where rental contracts are typically for a year) leads to a lag of up to a year in observed rental rates adjusting to pressures that affect rental rates.  As discussed in that post, this long lag has led to a divergence in observed inflation rates in the past year for the shelter component of the CPI in comparison to the CPI for all goods and services other than shelter.

Inflation in the shelter component of the CPI has been the primary cause of inflation remaining above the Fed’s 2% target.  Inflation in all goods and services in the CPI other than shelter moderated greatly in mid-2022 and has since fluctuated between zero and 2%.  But the shelter component of the CPI has kept the overall CPI at between 3 and 4% since mid-2022.  With the HICP leaving out the cost of shelter on owner-occupied homes (it includes it for those who rent), it is not surprising that inflation as measured by the HICP has been well below inflation as measured by the CPI.

b)  Also important to understanding the differing figures is that the HICP methodology does not include seasonal adjustments.  While seasonal factors can be important, adjusting the figures to reflect that seasonality is technically difficult.  The HICP methodology, as adopted by the EU, leaves it out.  This probably explains the low rates observed in the chart for HICP inflation seen in each of the six-month figures ending in December, with relatively high rates seen in each of the six-month figures ending in June.

Inflation as measured by the HICP will likely therefore go up in the coming months from the 0.0% rate observed for the six months ending in December 2023 and the 1.0% rate ending in January 2024.  Using a rolling 12-month average will mostly resolve such seasonality differences, and a chart of this will be examined below.  It shows 12-month rates for the HICP (both for the overall HICP and for a core HICP that leaves out food and energy) fluctuating around a 2% rate starting in June 2023 and continuing at least until now.

c)  There are a number of other technical differences, but these are likely less important for the issues being considered here.  For example, the HICP adjusts the weights used to calculate the overall HICP index (and its component sub-total indices) only once a year.  The CPI, in contrast, is what is called a chain-weighted index where the weights are changed each month to reflect changing expenditure shares.  But this is probably not terribly important as the weights do not change even year to year by all that much.

Also, the HICP – if one strictly followed the formal methodology – includes prices faced by the rural population.  But the BLS only collects price data from the major urban areas for the CPI, which means that the HICP for the US will only reflect urban prices.  That does, however, then mean that there will be less of a difference between the HICP as estimated for the US and the standard CPI for the US.  But it also then means comparisons of inflation across countries (where other countries include estimates for prices in rural areas) will not be as reliable.

Finally, the year-on-year inflation rates for the HIPC are of interest.  They have the advantage of mostly not being affected by seasonality issues (there can still be some seasonal effects, given how the seasonal adjustment algorithms work), but have the disadvantage of not capturing turning points in inflation trends as well.

The year-on-year rates for the US of both the overall HICP and the core HICP have been:

In terms of the year-on-year measures (12-month rolling changes ending on the dates shown), both the overall HICP and the core HICP have fluctuated at rates of between 1 1/2 and 2 1/2% since the 12-month period ending in June 2023.  It has remained within that narrow range for 8 months, or two-thirds of a year.  If the US measured inflation like Europe does, one would conclude that the Fed should now be allowing interest rates to fall from their current relatively high levels (aimed at reducing inflation) down to more neutral levels.

Inflation in the US as measured by the CPI remains above the Fed’s 2% target primarily due to inflation in the shelter component of the index.  But the behavior of the cost of shelter has been special.  This is in part due to the lag built into how the cost of shelter services is estimated for the CPI (due to reliance on estimates of rental-equivalent costs, as discussed in the post from last May cited above).  But there have also been other factors in recent years due to impacts arising from the response to the Covid crisis and then a rebound that came with the recovery from that crisis.  Those issues will be discussed in a subsequent post on this blog.

========================================================

Update – February 17, 2024

From comments I have received on this post, I see that some of the points should be clarified.  The basic message was clear enough:  That if the US were using the measure of inflation used in Europe, one would conclude that inflation is now at around the Fed taget of 2% and that the Fed should therefore consider allowing interest rates to fall back down to more neutral levels.  The primary reason why inflation in the US as measured by the CPI has remained above the Fed’s 2% target has been inflation in the cost of housing services (which is estimated based on the cost of rental equivalents).  And the estimates for housing are special, both due to lags in how rental rates are determined and special circumstances arising from the Covid crisis and the recovery from it.

The HICP measure used in Europe treats housing differently, as it measures the prices only of goods and services actually paid for, and not – as for owner-occupied homes – a service that follows from the ownership of the asset.  The US CPI treats the services of owner-occupied homes as if a rent were being paid to yourself, as the owner.  This leads to the not terribly intuitive situation where inflation in those implicit rents may be high – which is treated as if it were reducing your real income – but at the same time those rents are being paid to yourself – thus increasing your income.  That ends up as a wash.  But when we look at what has happened to real incomes as a consequence of inflation, we include the former (those implicit rents are reflected in the CPI) but leave out the latter (incomes are not adjusted for the implicit rents being paid to yourself).

In part for this reason, the European HICP measure of inflation leaves out those implicit rents.  But one should not say that the HICP is right and the CPI is wrong (or vice versa).  Rather, they are different and it is important to understand the difference.

In addition to the treatment of housing services, the HICP measure is not seasonally adjusted.  The CPI that is usually the focus of attention is seasonally adjusted.  I flagged this on the charts, but it likely would have been useful to have shown as well the not seasonally adjusted CPI series.  The charts become more cluttered, but one can then better see what the impact of seasonal adjustment (or the lack of it) has been.  And the not seasonally adjusted CPI is more directly comparable to the HICP.

The chart at the top of this post then becomes:

The rolling 6-month annualized change in the CPI is shown here as calculated both from the seasonally adjusted series (in red, and as before) and from the not seasonally adjusted series (in black, and with diamond markers).  As noted in the text, the seasonally adjusted CPI (in red) has fluctuated in the relatively narrow range of 3 to 4% (annualized) since the six-month period ending in December 2022 (i.e. since mid-2022).  The not seasonally adjusted CPI (in black) has moved on a similar path as the HICP (which is not seasonally adjusted), but always above it – and generally about 1 to 2% points above it (since mid-2022).

Adding the CPI and core CPI series to the 12-month rolling average chart might also have been helpful:

While 12-month changes do not capture the turning points as well as 6-month changes do, the basic story remains the same:  Inflation fell sharply in the 12 months leading up to June 2023 (i.e. since mid-2022), but the US CPI measure has been well above what the European HICP measure would indicate inflation has been over this recent period.  It is also interesting to note that while the overall CPI has been (since mid-2023) about 1 to 1 1/2% points above the comparable overall HICP measure, the core CPI has been about 2 to 2 1/2% points above the comparable core HICP measure.

Why?  Again this can be attributed to how the cost of housing services is treated. The HICP measure leaves out the implicit rents paid on owner-occupied housing, but the CPI includes them.  In the overall CPI, shelter has a weight of 36% in the overall index, which is significant.  Food and energy have a weight of a little over 20% in the overall index.  Food and energy are excluded from the core index, so the remaining items have a weight of about 80%.  The weight of shelter in the core CPI will therefore be 36% of 80% = 45%.  With the cost of housing services rising at a faster pace than the cost of other goods and services, the higher, 45%, weight of housing services in the core CPI leads to the margin over the core HICP (where services from owner-occupied housing are left out and hence have no weight) being greater.  Thus the 2 to 2 1/2% margin of the core CPI over the core HICP rather than the 1 to 1 1/2% margin of the overall CPI over the overall HICP.

 

Inflation Other Than for Shelter Has Moderated Even More – An Update With May 2023 Data

Posted on by

This is just a short update to my May 10 post on this blog to reflect newly released data.  The Bureau of Labor Statistics released today its estimates for the Consumer Price Index for May 2023.  I noted in the May 10 blog post (that had data through April 2023) that the pace of inflation when one excludes the shelter component had come down sharply over the past year.  One sees this most clearly when one focuses on the change over rolling 6-month periods (annualized). The rolling 6-month change in the CPI excluding the shelter component peaked in early 2022 at an annualized rate of over 12%.  But since late 2022, the 6-month rate has fluctuated in the range of just 0.3 to 2.0% (annualized), and it remains there.  The May 2023 6-month figure was just 0.8%.  The Fed’s target is to keep inflation at around 2%.  These have been below that for half a year now.

The most interesting new data point in the chart above is the most recent 12-month change in the CPI excluding the shelter component.  It had hit a peak of over 10 1/2% in mid-2022, but had come down to a rate of 3.4% in March as well as in April.  In the newly released data as of May, it had fallen further to just 2.2% on a 12-month basis.  This is basically at the Fed target for inflation.

But the shelter component of the CPI of course matters.  It has a 35% weight in the overall CPI index as it includes not only what people pay when they rent housing but also the rental equivalent cost of owner-occupied housing (which can only be estimated by observing what is being paid for rental units).  As was explained in the May 10 post on this blog, the shelter component of the CPI can furthermore only be estimated by observations of what people are actually paying as they rent, and rental contracts in the US are normally set for a year.  Thus even where there may be pressures to increase rental rates in some market, leading to higher rental rates being charged as contracts come up for renewal, those higher rents will only go into effect when rental contracts are in fact renewed.  If you had renewed your rental contract recently, it might be close to a year before the rent you have to pay actually goes up.  Since the Bureau of Labor Statistics interviewers ask the households what they are actually paying in rent at the time of the interview, inflationary pressures on rental rates will take up to a year to work their way through.

This long lag is seen in the orange and red lines in the chart above.  The annualized rates rose throughout 2021 and 2022.  But the 6-month rate peaked in early 2023, reaching a rate of 9.1% (annualized) as of February.  As of May it has come down to 7.9%.  The 12-month rate peaked at 8.2% in March and is now down slightly to 8.0%.  While the news media and others normally focus on the 12-month rates, turning points will often be first revealed by examining shorter time periods.  If too short (such as monthly), it may be difficult to isolate the trends from the statistical noise inherent in the monthly data.  A 6-month rate is a reasonable compromise.

The cost of housing is still rising at too high a rate.  But given the 12-month periods of most rental contracts, and the use of such observed rental rates to impute the rental-equivalent costs of owner-occupied housing, it will take some time for the shelter component of the CPI to return to the rates observed prior to the onset of the Covid crisis.  The inflation rates are now coming down, but that cycle is not yet complete.

But excluding shelter, consumer inflation is already back to where it was before the disruptions due to the Covid crisis, with its lockdowns, supply-chain disruptions, and the massive fiscal relief packages passed into law under both Trump and Biden.  Hopefully the Fed is paying attention to this.