All Federal Reserve Banks operate within a specific geographic area—called a _district_. The Cleveland Fed is in the Fourth District of the Federal Reserve System, which covers Ohio and parts of Pennsylvania, Kentucky, and West Virginia. The Cleveland Fed’s Survey of Regional Conditions and Expectations (SORCE) covers the Fourth District, and FRED now includes 27 of those data series. The FRED graph above shows the broadest summary of business conditions captured by SORCE. * Total (solid blue line) * Manufacturing and transportation (dashed green line) * Other (dashed orange line) The latest value for total business conditions at the time of this writing is 8. And what does that mean? Each index value shows the difference between the percentage of survey respondents reporting increases and the percentage reporting decreases in business conditions, including demand, employment, inflation, etc. Measures calculated this way are often called _diffusion indexes_. These values vary from positive to negative: The index shows a negative value when survey respondents reporting decreases outweigh survey respondents reporting increases. It shows a positive value when survey respondents reporting increases outweigh survey respondents reporting decreases. Between June 2014 and the time of this writing, the SORCE index values have ranged between a maximum of 67 (May 2018) and a minimum of -80 (May 2020). Most frequently, they oscillate between 30 and -20. FRED has an array of regional Fed datasets, with a few of them listed below. * **Chicago Fed’s**Midwest Economy Index * **Dallas Fed’s**Texas Employment Data * **New York Fed’s**Empire State Manufacturing Survey **How this graph was created** : Search FRED for and select “Total Business Conditions, Diffusion Index for Survey of Regional Conditions and Expectations (SORCE).” Click on the “Edit Graph” button and select the “Add Line” tab to search for and select “Manufacturing & Transportation Business Conditions, Diffusion Index for Survey of Regional Conditions and Expectations (SORCE).” Don’t forget to click on “Add data series.” Repeat the last two steps to search for and add “Other Industries Business Conditions, Diffusion Index for Survey of Regional Conditions and Expectations (SORCE).” Suggested by Diego Mendez-Carbajo.

What’s really happening in US manufacturing? In our FRED graph above, we compare labor productivity for all workers in the manufacturing sector alongside the share of manufacturing employment in total US nonfarm employment since 1987. The graph tells a compelling story: We’re producing more per worker in manufacturing, but with far fewer workers relative to the overall workforce. Manufacturing productivity rose strongly from the late 1980s until the Great Recession in 2008-09. During this period, workers were producing more value per hour due to advances in automation, more-efficient production processes, and improved capital equipment. At the same time, the share of manufacturing workers in the total nonfarm workforce has been shrinking. This shows that even as factories are producing more, they require a smaller share of the labor force. Notably, these trends—often refereed to as structural transformation—were underway long before the China trade shock of the early 2000s. Since 2010 or so, we see that labor productivity in manufacturing has largely flattened, showing little growth compared with the period before 2010. Meanwhile, the decline in the manufacturing employment share has also slowed considerably. This divergence has large implications. First, it underscores how innovation is reshaping manufacturing. The trends displayed above are consistent with ideas surrounding automation and technology adoption. Second, it challenges notions about manufacturing being a major job engine. If productivity keeps rising while the share of manufacturing employment falls, employment gains in the economy will need to come from other sectors. **How this graph was created** : Search FRED for “All Employees, Manufacturing.” Above the graph on the right, click on “Edit Graph,” add a series by searching for “All Employees, Total Nonfarm,” and apply formula _a/b_. Open the “Add Line” tab and search for and select “Manufacturing Sector: Labor Productivity (Output per Hour) for All Workers.” Open the “Format” tab and place the legend on the right for the second line. Start the graph on 1987-01-01. Suggested by Alexander Bick and Kevin Bloodworth II.

In 2025, many US households reported a surge in electricity prices. According to the producer price index (PPI) for residential electric power, these prices have gone up by 7% between January and August 2025. This rise in prices can be attributed to many supply and demand factors, including an aging power infrastructure and the construction of new data centers throughout the US, which are a major contributor to greater demand for electricity. **Looking at the data** Our FRED graph above shows the PPI values for both natural gas and residential electric power. Since 2015, the two have generally moved together. What stands out immediately is the sharp divergence between natural gas prices and electricity prices starting around 2023. Natural gas has long been a core input for US power generation. As growth in supply outpaced demand, natural gas prices dropped significantly. Under normal circumstances, falling natural gas prices might ease wholesale electricity costs, but that relationship appears increasingly muted. The steeper increases in residential electricity prices in recent months reflect structural pressures in the power sector that go beyond fuel costs. These include the need for extensive grid and transmission upgrades, rising operational costs, and surging electricity demand from large consumers such as data centers and other power-intensive facilities. This widening gap hints at an increasingly strained electricity system, where capital investment needs and surging industrial demand are outpacing the benefits of cheaper natural gas. If these trends continue, electricity prices may remain elevated even in an environment of low fuel costs, reshaping energy planning for households, businesses, and policymakers. **How this graph was created** : Search FRED for and select “PPI Natural gas.” Above the graph, click “Edit Graph,” open the “Add Line” tab, and search for and add “PPI Residential electric.” Open the “Format” tab and place the legend of the second line to the right. Start the sample period on 2015-12-01. Suggested by Alexander Bick and Kevin Bloodworth II.

Central index of economics institutions (academic, governmental and non-profit) organized by country and US state.

Gross domestic product (GDP) is the sum of all incomes distributed in a year, including disposable personal income (DPI). So, GDP and DPI are positively correlated. That is, they generally move in the same direction and follow each other closely. Our FRED graph above shows this correlation, with one striking deviation at the point of the COVID-19 pandemic. What’s behind this deviation? A third measure, personal current transfer receipts, can help explain. The US government increased transfers to households in such large proportions during the pandemic that DPI growth actually increased even though GDP growth decreased. **GDP and DPI** Prior to 2020, GDP and DPI show a clear positive correlation. During recessions (shaded areas in the graph), the growth rates of GDP and DPI decreased. The correlation also holds outside recessions: During the recoveries after the 1981-82 and 2007-09 recessions, the growth rates of both GDP and DPI increased. From 2020 to 2023, however, the two series behave quite differently: GDP’s growth rate decreased while DPI’s growth rate increased. **DPI and Transfers** Although GDP and DPI are closely related, there’s a difference between them: DPI includes _government transfers_ to individuals, which is captured by the thin red line in the graph. And it’s these transfers that explain the change in the behavior of GDP and DPI in 2020. Our second FRED graph, above, zooms in on the period of the COVID-19 pandemic. The US government sent transfer payments to households to alleviate the severity of the crisis. In the second quarter of 2020, transfers increased substantially and so did personal disposable income, despite the fact that GDP growth had gone negative. The growth rates of transfers and PDI then generally declined and went negative in the first quarter of 2022, despite the fact that GDP growth had been positive. By 2024, the rates were aligned once again. **Why was 2020-23** **so different**? The US government has increased transfers to households before to alleviate the severity of crises. For example, the first graph shows an increase in transfers during the 2007-09 recession. So why did DPI growth actually increase during the COVID-related recession, but in no other recession? The increase in transfers in 2008 was only 25%, compared with 75% in 2020 and 89% in 2021. Those are proportions not seen before. **How these graphs were created** : Search FRED for and select “GDP.” Click on “Edit Graph,” open the “Add Line” tab, and search for and select “DPI.” Add another line search for “PCTR.” Open the “Edit Lines” tab and choose quarterly frequency and “Percent Change from Year Ago” units for each line. From the “Format” tab, choose the right axis for the third line (PCTR). For the second graph, start the sample period in 2020 Q1. Suggested by Guillaume Vandenbroucke.

RePEc is a central index of economics research, including working papers, articles and software code

RePEc is a central index of economics research, including working papers, articles and software code

The current softening in the labor market is hitting recent college graduates especially hard, suggesting the traditional college premium may be weakening. At least for quickly landing a job. In this post, we take a longer view of the unemployment gap between college graduates (with a bachelor’s degree or higher) and high school graduates (with no college). The graph shows a persistent and significant disparity between the two unemployment rates: From 2000 to 2025, high school graduates consistently faced unemployment rates at least 2.3 percentage points higher than those for college graduates. This enduring gap reflects structural differences in the _types_ of jobs each group holds: College-educated workers are more likely to have jobs that are less susceptible to cyclical layoffs and economic disruptions. **The gap is most pronounced during economic downturns.** During the Great Recession (2008-2010), the unemployment rate gap between workers with and without college degrees spiked dramatically, to about 7.8 percentage points. We see a similar pattern during the COVID-19 pandemic in 2020. This pattern highlights the greater vulnerability that less-educated workers have to economic shocks and suggests that higher education provides both access to better employment opportunities and greater job security during recessions. **The gap shrinks but persists in times of tight labor markets.** In tight labor markets, the gap narrows significantly but never disappears. During the long recovery of the 2010s, as the labor market improved, the gap gradually shrank to just above 2 percentage points. A recent FRED blog post shows that the unemployment rate gap for _young workers_ disappeared altogether during the first months of post-pandemic recovery and has remained at historically low levels since then. **Higher unemployment can have lasting effects.** The lack of job security for workers without college experience can lead to lower wages and lower lifetime earnings, as frequent job losses interrupt career progression and skill development. This educational divide in employment outcomes underscores the economic value of higher education in the American labor market. **How this graph was created** : With data graphing tools in FRED, we’re able to subtract the unemployment rate of college graduates from the unemployment rate of high school graduates and graph a single data series. Search FRED for and select “Unemployment Rate – College Graduates – Bachelor’s Degree and Higher, 16 years and over.” From the “Edit Graph” panel, use the “Customize data” field to search for and add “Unemployment Rate – High School Graduates, No College, 16 years and over.” In the formula section, type _b – a_. Under “Modify Frequency,” choose “Semiannually” and keep the “Aggregation Method” as “Average.” Suggested by Serdar Ozkan and Nicholas Sullivan.

RePEc is a central index of economics research, including working papers, articles and software code

**How do prices change over time?** Prices tend to go up, especially over long periods. But some prices can increase much faster than others and some can even decrease. Over the past 40 years or so, price changes for goods have tended to be lower than price changes for services. Our FRED graph above uses data from the consumer price index release table to track price changes for some specific **goods**. The average CPI, shown by the solid blue line, is noticeably higher than the values for all the specific goods listed in the graph. Our FRED graph below tracks price changes for some specific **services**. The average CPI, shown by the same solid blue line, is noticeably lower than the values for all the services listed in the graph. **W****hy are price changes for goods below average?** Increases in quality and productivity have allowed for slower increases in the prices of many goods and even some actual price declines. The quality of cars has improved considerably. These enhancements have to be factored-out to get a genuine apples-to-apples price reading that can be compared over time. For example: If you pay $35,000 for a car today and 8 years ago you paid $25,000, then the difference in purchase price is $10,000. But consider this: You’re also getting the most modern features with your new car, such as automatic emergency braking and keyless entry with biometric access. Those features weren’t available when you bought your previous car. So, because you’re getting much greater value with this purchase, the adjusted CPI price increase would be much less than $10,000. The same applies to computer equipment, which wasn’t even a consumer good in the early 1980s. Productivity in manufacturing has also improved considerably. Apparel, toys, and car parts, for example, can be produced much more efficiently or in lower-cost locations than before. As engineers find more efficient ways of making things, those things become (relatively) cheaper. **Why are price changes for services above average?** It is much more difficult to improve productivity for services than it is for goods, so services have gotten much more expensive. Consider education and medical care, for example. You cannot successfully increase classroom size beyond some point, and you cannot put more patients in a single hospital bed. Another example is the barber: Technology improvements in hair cutting have been very limited, and barbers still serve roughly the same number of customers each day as they did 40 years ago. Of course, the rise of artificial intelligence may create new trends for some services. A topic for another day. **One more thing to know about the CPI** A good that has actually gotten much more expensive over time is tobacco products. These products have had increasingly higher taxation for several decades, with the intent to reduce their consumption. Helpfully, the CPI accounts for taxes when it calculates price changes. **How these graphs were created:** Start from the CPI release table noted above, which you can find in the notes for the CPI graph or any of its components in FRED. Check the series you want displayed. Click on “Add to graph.” Change the start year to 1983. Suggested by Christian Zimmermann.