1/12/22

Переоценка инфляции в США с поправкой на изменение активов ФРС с 2010 по 2021 гг.

Девальвация валюты — это процесс, отличный от инфляции цен, которая обусловлена только внутриэкономическими факторами. "Вертолетные" деньги, вливаемые в экономику, эквивалентны девальвации: человек получает больше единиц платежа без соответствующего изменения количества товаров и услуг для покупки. Федеральная резервная система США вводила в экономику деньги через различные механизмы с 2009 года, как показано на рисунке 1 (заимствовано с веб-сайта Federal Reserve). Можно видеть, что всплеск 2020 года был чрезвычайно большим как взрыв. Мы уже обсудили этот эффект и описали его влияние на текущие оценки инфляции в этом блоге. Мы ожидаем ценовой дефляции во второй половине 2022 года. 

Модель, связывающая скорость изменения реального ВВП на душу населения, rGDPpc, и уровня безработицы, u, критически зависит от точности оценок обоих параметров. Реакция на меры Федеральной резервной системы в период с 2009 по 2020 год должна быть аналогична реакции на рост активов в 2020 и 2021 годах — девальвация, проявляющаяся в виде ценовой инфляции. Разницу между реальным экономическим поведением и денежными спекуляциями следует рассматривать в долгосрочной эволюции экономической связи между rGDPpc и u. Как видно из рис. 1, фактор девальвации до 2009 г. отсутствовал. Таким образом, введение столь значимого фактора в основные экономические показатели (влияние совокупного количества денег на номинальный ВВП) должно было проявляться в виде структурного разрыва в долговременной функциональной связи rGDPpc и u. Такой разрыв мы нашли в 2010 г. – коэффициент линейной регрессии составлял -0,465 до 2010 г. и только -0,26 после. Эти оценки были сделаны методами, основанными на среднеквадратичных ошибках. Константа регрессии также изменилась с 0,907 до -0,25. Изменение этих коэффициентов позволило сопоставить данные за период с 2010 по 2019 год, как показано на рисунке 2. Коэффициенты регрессии, рассчитанные для периода между 1979 г. (год значительного пересмотра определения ВВП и безработицы) и 2009 г. (показаны красным), не могут предсказать уровень безработицы на основе реального ВВП на душу населения. Тем не менее, даже обновленная модель выходит из строя в 2020 году, о чем говорится в этом посте.

 

Учитывая (вертолетные) деньги, влитые в экономику США, мы предполагаем, что экономически обоснованные оценки реального ВВП на душу населения были искажены девальвацией, добавленной к фактической инфляции цен. Эта гипотеза предполагает, что реальные значения ВВП были занижены, поскольку уровень инфляции в США был завышен. При компенсации на девальвацию коэффициенты регрессии, полученные за период с 1979 по 2009 год, должны быть действительными, и мы можем оценить rGDPpc из u, используя долгосрочную связь. Чтобы спрогнозировать правильный уровень инфляции и, следовательно, реальный ВВП, мы ввели линейную функцию инфляции, изменяющуюся с 3,8% в 2010 г. до 2,8% в 2019 г. - впрыск денег не был постоянным во времени. На рис. 3 сравниваются опубликованные и скорректированные оценки rGDPpc, а на рис. 4 представлено соответствие модели между 2010 и 2019 годами с исходными коэффициентами регрессии. На рисунке 5 сравниваются скорости изменения rGDPpc для скорректированных и опубликованных значений. Фактические темпы инфляции цен были ниже опубликованных на разницу между скорректированными и опубликованными значениями rGDPpc, как показано на рисунке 6. 

Поправка инфляции цен на девальвацию, вызванную действиями ФРС,  предполагает, что реальный ВВП на душу населения рос в среднем на 3,2% в год в период с 2010 по 2019 год вместо опубликованного значения (в среднем) 1,5% в год за тот же период. Это результат денежной поддержки Федеральной резервной системы. Эта разница в скорости изменения в 1,7% соответствует общей сумме активов ФРС: ВВП  примерно 20 трлн в год, умноженный на 0,017 = 0,34 трлн в год, или 3,4 трлн всего в период с 2010 по 2019 год. Это означает, что все деньги, влитые ФРС в экономику, в нее попали и вызвали общее повышение цен ровно на сумму повышения активов ФРС. Таким образом связь rGDPpc-u в период между 1979 и 2009 годами все еще работает при очистке данных по инфляции. На рис. 4 показана измеренная кривая уровня безработицы на основе этой связи. Для прогнозируемой кривой реальный ВВП на душу населения должен упасть на 8,1% в 2020 г. и увеличиться на 7,5% в 2021 г. На рис. 4 представлена ​​ оценка на 2021 г.: 67200 долл. США.

  

Рисунок 1. Активы Федеральной резервной системы  


Рис. 2. После 2010 г. связь между rGDPpc и u с обновленными коэффициентами регрессии (черный) и с коэффициентами регрессии, полученными за период с 1979 по 2009 г. (красный). До 2010 г. исходный временной ряд представлен черными точками. 

 


Рисунок 3. Опубликованный (красный) и скорректированный (черный) реальный ВВП на душу населения в период с 2010 по 2020 год. Модельная оценка на 2021 год составляет 67200 долларов США. 

 


Рисунок 4. Соответствие между измеренным и прогнозируемым уровнем безработицы для скорректированного rGDPpc. 

 

Рисунок 5. Скорость изменения rGDPpc для опубликованных и скорректированных оценок.

  

Рисунок 6. Разница между опубликованным и скорректированным уровнем инфляции, полученная из рисунка 5.



Re-estimation of price inflation in the US by correction for the change in the Federal Reserve's assets from 2010 to 2021

 Currency devaluation is a process different from price inflation driven by economic factors. The helicopter money poured into an economy is an equivalent of devaluation – one gets more units of payment without a corresponding change in the amount of goods and services to buy. The US Federal Reserve inserted money through various mechanisms since 2009 as Figure 1 shows (borrowed from the Reds website). One can see that the 2020 surge was extremely large like an explosion. We discussed this effect and described its effect at the current inflation estimates. We are expecting price deflation in the second half of 2022. 

The model linking the change rate of the real GDP per capita, rGDPpc, and of the unemployment rate, u, is critically dependent on the estimates of both parameters. The reaction to the measures of the Federal Reserve between 2009 and 2020 should be similar to that to the asset growth in 2020 and 2021 – devaluation measured as price inflation. The difference between real economic behavior and money speculations should be seen in the long-term evolution of the economic link between rGDPpc and u. There was no devaluation factor before 2009. Thus the introduction of such a significant factor into major economic measures (aggregate money influence on nominal GDP) should be manifested as a structural break in the rGDPpc and u link. We have found such a break in 2010 – the coefficient of linear regression was -0.465 before 2010 and -0.26 after. These estimates were made by the RMS-based methods. The regression constant also changed from 0.907 to -0.25. This allowed fitting the data between 2010 and 2019 as Figure 2 shows. The regression coefficients estimated for the period between 1979 (major definitional revision to GDP and unemployment) and 2009 (shown in red) fail to predict the rate of unemployment from the real GDP per capita. Nevertheless, even the updated model fails in 2020, as discussed in this post. 

Considering the (helicopter) money injected into the US economy we suggest that the economically based estimates of real GDP per capita were distorted by devaluation added to actual economically driven price inflation. This hypothesis suggests that the real GDP values were underestimated as the US inflation rate was overestimated. In this case, the regression coefficients obtained for the period 1979 to 2009 should be valid and we can estimate rGDPpc from u using the long-term link. To predict the correct inflation rate and thus the real GDP we introduced a linear function changing from 3.8% in 2010 to 2.8% in 2019. This is needed to match the Federal Reserve history in asset change during this period (Figure 1) as the money injection was not constant in time. Figure 3 compares the published and corrected rGDPpc estimates and Figure 4 presented the model fit between 2010 and 2019 with the original regression coefficients. Figure 5 compares the change rates in the rGDPpc for the corrected and published values. The actual rate of price inflation was lower than published by the difference between the corrected and published rGDPpc values as Figure 6 illustrates. 

The price inflation correction suggests that the real GDP per capita has been growing at an average rate of 3.2% per year between 2010 and 2019 instead of the published value of 1.5% per year on average for the same period. This is the result of the Federal Reserve's money support. This 1.7% change rate difference fits the total amount of assets: the GDP of 20 trillion per year times 0.017 = 0.34 trillion per year or 3.4 trillion in total between 2010 and 2019.  The injected money added its fake value to the economically based price inflation. The rGDPpc-u link between 1979 and 2009 is still working when cleaning the inflation data. Figure 4 shows the unemployment rate predicted curve based on this link. For the predicted curve, the real GDP per capita has to drop by 8.1% in 2020 and increase by 7.5% in 2021. Figure 4 presents an estimate for 2021: $67,200.

 

Figure 1. Federal Reserve assets

 

Figure 2. After 2010, the link between rGDPpc and u with the updated regression coefficients (black) and with the regression coefficients obtained for the period between 1979 and 2009 (red). Before 2010, the original time series is presented by black dots.  

 

 Figure 3. The published (red) and corrected (black) real GDP per capita between 2010 and 2020.  For 2021, the model estimate is $67,200.

Figure 4. The fit between the measured and predicted rate of unemployment for the corrected rGDPpc.

 

Figure 5. The change rate in the rGDPpc for the published and corrected estimates.

 

Figure 6. The difference between the published and corrected inflation rate as obtained from Figure 5.

 

 

 

Helicopter money distort the GDP estimates in the USA

In our previous post, we described the difference between the (BEA) published and predicted by our model values of the real GDP per capita, rGDPpc. We have developed and tested since 2005 a model linking the change rate in rGDPpc, and the change rate of the unemployment rate, u (paper here): 

            du = a + bd[ln(rGDPpc)] , 

where a and b are regression coefficients, which depend on the change in measurement units. This difference between the published and predicted values was explained by the government transfer in 2020 and 2021 which has never been observed before. This was “helicopter” money taken from some sources out of the US economy, i.e. not generated in any economic activity, and considered as a part of the economy. In this post, we describe the unique properties of this transfer and explain the conflict between rGDPpc and u.   

One of the largest components of the (nominal) GDP is Personal Income, PI, which is published by the BEA. This parameter served as the main channel to pour the “helicopter” money into the US economy in 2020 and 2021. Among its constituent components, the PI contains “Personal current transfer receipts”, PCTR, which is a parameter including “Government social benefits to persons” as a major component. Figure 1 presents the evolution of the PI and PCTR since 1947. (Please notice the logarithmic income scale.) Figure 2 presents the evolution of these two parameters normalized to their respective levels in 1947. The PCTR share in the PI has been growing much faster than the PI itself. The surge in the PCTR’s share published for 2020 and 2021 has never been observed during the measurement period. One can also notice that the PI curve had a short-term fall in 2009 accompanied by a significant rise in the PCTR curve. This effect has a clear explanation – the unemployment benefits (a part of the PCTR) compensated for the surge in unemployment. In 2020 and 2021, there is no fall in the PI curve which is accompanied by a dramatic surge in the PCTR. This pattern means that the fall in the PI induced by a dramatic increase in the unemployment rate was overcompensated by the surge in PCTR. The PI in 2021 was 105% of the GDP in 2020! If real, such a configuration of the PI and GDP would mean the economic "Perpetual Motion Machine". 

Figure 3 presents the PCTR/PI ratio to illustrate the growth in the PCTR share since 1947. The spike in 2020 and 2021 is unprecedented. In 2008, the growth in PCTR was much smaller. Figure 4 stresses the difference between the government reaction in 2009 and 2020. The PI fell by approximately 400 billion dollars in 2009 as a reaction to the Great Recession. The PCTR was increased by approximately 100 billion dollars in 2008 and 2009 relative to the 2007 level. This was an economic reaction to the recession despite the dangerous situation later called the Great Recession. The growth between 2010 and 2019 was not accompanied by an increase in the PCTR. There were other channels to pump money into the economy and galvanize it. In 2020, one could expect a much deeper fall in the PI than that in 2009. The PCTR was used to avoid such a fall and one can see that the PI annual increment grew instead. The same situation was observed in 2021. 

In 2008, the rise had economic reasons and was accompanied by a drop in real GDP per capita described by a long-term linear relationship. In 2020, the surge in u was non-economic and was not accompanied by a proportional fall in the real GDP per capita. The difference in reaction is explained by the government transfer of a few trillion dollars. The decrease in u between 2009 and 2019 was driven by economic growth. The drop in 2021 is also caused by a 9% spike in real GDP per capita.

  

Figure 1. The evolution of Personal Income, PI, and Personal Current Transfer Receipts, PCTR: 1947 to 2021. Notice logarithmic scale.

Figure 2. The evolution of Personal Income and Personal Current Transfer Receipts normalized to their respective values in 1947. Notice the surge in 2020 and 2021. 

Figure 3. The evolution of PCTR/PI ratio 

Figure 4. Annual increment of the PI and PCTR. Notice the difference in reaction in 2008 and 2020. 

Figure 5. The rate of unemployment between 1947 and 2021. The largest annual rises were observed in 2008 (to 9.3 from 5.8 in 2007) and in 2020 (to 8.1 from 3.7 in 2019).

 

1/11/22

The rate of unemployment in the USA in 2020 and 2021 indicates the the real GDP per capita in 2021 is only 87.5% of that in 2019

 We have developed and published (last version in 2021) an economic model linking the change rate in the real GDP per capita and the rate of unemployment. Figure 1 presents the performance of this model since 1947 when the real GDP and unemployment estimates became more or less reliable. There are four structural breaks in the model due to severe changes in the real GDP (e.g., imputed rent introduction) and unemployment estimates. These breaks are introduced in the years of the definitional changes, not from data behavior, and thus are legitimate. Statistical estimates (regression and the error) of the model for the period between 1947 and 2019 are presented in Figures 2 and 3. Overall, the model is excellent in the prediction of unemployment from the real GDP per capita and vice versa. We do not discuss the causality direction here, since both parameters depend on more fundamental factors. Essentially, the real GDP and unemployment are different measures of economic growth. 

The current pandemic ruins the model, as Figure 3 shows. The rate of unemployment (data from bls.gov) is much higher than that predicted by the real GDP per capita change (data from bea.gov). For 2020, the unemployment rate was 8.1% and the prediction was only 4%. The model failure may have different causes, and we assume that the estimates of real GDP per capita are wrong because it includes the government transfer taken from sources outside the US economy. The COVID-19 driven money transfer is counted as an internal economic source and passes to the real GDP estimates through the Personal Consumer Expenditures. Since the causality direction is not applicable to the model, we have two estimates of the same economic parameter driving both measures, GDP, and unemployment. Unemployment is easier to count, and the estimate of 8.1% is rather accurate for 2020. The real GDP per capita corresponding to this rate has to drop by 19.8% from the 2019 level. In this blog, we made a similar estimate of a 21.7% drop in real GDP a year ago. The reason for the difference between the published and predicted real GDP per capita was the same – government transfers. Now, the estimate of the unemployment rate supports our previous finding.  Figure 4 shows the same curves as in Figure 1, but the real GDP estimates are calculated from the observed unemployment rates in 2020 and 2021. The drop in real GDP per capita in 2020 was 19.8% and the growth in 2021 – 9%. These estimates mean that, in real terms, the US economy is currently at the level of 87.5% of that in 2019. The real economic fall since 2019 is 12.5%

 

Figure 1. The observed and predicted from real GDP per capita rate of unemployment.

 

Figure 2. Statistical link between the measured and predicted rates of unemployment between 1947 and 2019.

 

Figure 3. Time evolution of the model error between 1947 and 2019

 


Figure 4. Same as in Figure 1 for the period 2005 to 2021 and the real GDP estimates obtained from the rate of unemployment.

1/10/22

Employment-population ratio in the USA - far from a healthy state

The long-term decrease in the labor force participation rate described in this post is accompanied (and is a principal part of) by the drop in the employment-population ratio. Figure 1 presents the most recent data and two trends from 1975 to 2000 and from 2000 to 2025. Two drops in 2009 and 2020 are the most prominent. The workforce leakage is a challenge to the US economy: 1% drop is equivalent to 2,600,000. The drop by 7.5% in 25 years is equivalent to 18,000,000 people leaving the workforce. For me, it is not clear what is behind such an exodus. As mentioned before - older people (50+) are rather increasing their participation.  Younger people's failure to get jobs in the same proportion as in the past is a clear economic symptom  as based on the increasing difficulties in personal income growth. 


Figure 1. employment-population ratio in the USA


The labor force in the USA: participation of older people rising – younger on the fall

The labor force participation rate, LFPR, is on a long-term decline in the USA. Figure 1 demonstrates that the bottom value of approximately 58% will be reached around 2030. In 2021, the LFPR was at 61.7%. The red curve is the time inverted original (black) curve. For the periodic evolution, such an inversion can predict the future. We assumed the periodicity fifteen years ago and this prediction is still working accurately. 

At the same time, one can split the overall decline into the age-dependent groups as Figure 2 shows. The most vulnerable fractions of the population are younger people, especially the youngest (16-17). The older groups (55-59 and 60-64)  partially compensate for the drop in the younger groups and their LFPR has been growing since 2000, i.e. from the start of the long-term decline (Figure 1). The mid-age groups are not suffering any big changes in the LFPR.  


Figure 1.


Figure 2.

1/4/22

Productivity in manufacturing fails to show any growth in the USA since 2001

 In the USA, productivity in manufacturing in 2021 is the same as in 2001. In business, productivity growth is fast. The reason behind such a deviation is not clear. Could it be insufficient funding of industry? Something like "defund the industry"?





Переоценка инфляции в США с поправкой на изменение активов ФРС с 2010 по 2021 гг.

Девальвация валюты — это процесс, отличный от инфляции цен, которая обусловлена только внутриэкономическими факторами. "Вертолетные&quo...