Use tsa.filters.hp filter to calculate output-gap in Python
Enter a time series for GDP, industrial production, or another output measure. This calculator applies the Hodrick-Prescott filter logic used in Python workflows and estimates the trend component and output gap for each period.
Results
Run the calculator to see the latest trend estimate, cyclical deviation, and the full output gap chart.
How to use tsa.filters.hp filter to calculate output-gap in Python
If you want to use tsa.filters.hp filter to calculate output-gap in Python, you are really trying to split an observed economic series into two parts: a smooth long-run trend and a cyclical component around that trend. The cyclical component is commonly interpreted as the output gap, especially when the series is real GDP or an index of production. In practical macroeconomic analysis, this is one of the fastest ways to estimate whether the economy is running above trend, below trend, or roughly in line with potential.
In the Python ecosystem, the most common implementation comes from statsmodels. Many analysts call it through statsmodels.api as sm and then use sm.tsa.filters.hpfilter(...). Others import directly from the filter module. In either case, the idea is the same: the HP filter chooses a trend that balances goodness of fit against smoothness. The stronger the smoothing parameter, called lambda, the smoother the trend becomes.
The calculator above gives you the same conceptual workflow inside a browser. You provide a numeric series, choose a frequency, and calculate the filtered trend and gap. If you use a log transformation, the cyclical component is especially easy to interpret because it approximates a percent deviation from trend. That is why macroeconomists often work with the natural log of GDP before applying the filter.
What the output gap means
The output gap is the difference between actual output and trend or potential output. A positive gap suggests the economy is operating above its sustainable trend, which can be associated with inflation pressure, capacity constraints, or unusually strong demand. A negative gap indicates underutilized resources, weak demand, or recessionary conditions. Although the HP filter does not directly estimate structural potential output in the same way a full production-function model might, it provides a practical reduced-form estimate that is useful for exploratory analysis, charting, and forecasting diagnostics.
The HP filter formula in plain language
The Hodrick-Prescott filter solves an optimization problem. It selects a trend series that stays close to the observed data while also penalizing changes in the second difference of the trend. In other words, it does not want the trend to wiggle too much. The tradeoff is controlled by lambda.
- A lower lambda follows the data more closely and produces a less smooth trend.
- A higher lambda produces a smoother trend and larger cyclical swings.
- Frequency matters because quarterly and monthly data are naturally noisier than annual data.
The most widely cited settings are based on standard macroeconomic practice: 6.25 for annual data, 1600 for quarterly data, and 129600 for monthly data. These values are not arbitrary. They reflect the convention that smoothing should rise sharply as data are sampled more frequently.
| Data frequency | Common lambda | Typical use case | Interpretation |
|---|---|---|---|
| Annual | 6.25 | Long historical GDP or productivity series | Very limited number of observations, less aggressive smoothing needed |
| Quarterly | 1600 | Standard macroeconomic business-cycle analysis | Most common choice for real GDP, output, or aggregate demand studies |
| Monthly | 129600 | Industrial production, employment, or high-frequency activity data | Much smoother trend because monthly data contain more short-run noise |
Python example using statsmodels
In Python, the usual pattern is straightforward. First, import your data into a pandas Series. Second, decide whether to work in levels or logs. Third, apply the HP filter. Finally, calculate the output gap either as the raw cyclical component or as a percent deviation from trend.
import numpy as np
import pandas as pd
import statsmodels.api as sm
# Example series: real GDP
gdp = pd.Series([19800, 19870, 19910, 19980, 20020, 20090, 20140, 20210])
# Common practice: use logs for easier interpretation
log_gdp = np.log(gdp)
cycle, trend = sm.tsa.filters.hpfilter(log_gdp, lamb=1600)
# Approximate percent output gap
output_gap_pct = 100 * cycle
result = pd.DataFrame({
"gdp": gdp,
"log_gdp": log_gdp,
"trend_log": trend,
"output_gap_pct": output_gap_pct
})
print(result.tail())
A lot of users search for the phrase use tsa.filters.hp filter to calculate output-gap in Python because they already know the command family but want to connect the code with the economic interpretation. The key line is the one that multiplies the cyclical component by 100 after filtering logged data. That is the step that turns the cycle into an intuitive gap measure.
Why many analysts log the data first
Applying the HP filter to logged GDP has two advantages. First, the variance of macroeconomic series often grows with the level of the series, so logs can stabilize the scale. Second, the cyclical component becomes approximately interpretable in percentage terms. For example, a cycle value of 0.012 means actual output is about 1.2 percent above trend. This is more useful in macroeconomic communication than saying output is 250 chained dollars above a fitted trend, especially when data are revised over time.
If your series contains zeros or negative values, do not log it. In that case, you can filter the level directly and report the gap either in raw units or as a ratio relative to trend. The calculator above supports both options.
Interpreting the results from the calculator
- Actual value shows the latest observation in your series.
- Trend value is the smoothed level produced by the HP filter.
- Absolute gap is actual minus trend in raw units.
- Output gap is the percent deviation from trend, or the logged cyclical component times 100.
On the chart, the line series display actual output and trend output. The bar series displays the output gap percentage. This mixed view is useful because it lets you see whether a positive gap comes from a short-lived burst in activity or from a longer run acceleration that may eventually shift the estimated trend itself.
Real macro statistics that help contextualize output-gap analysis
Output-gap work makes the most sense when connected to real macroeconomic indicators. In the United States, real GDP growth from the Bureau of Economic Analysis and unemployment from the Bureau of Labor Statistics are two natural anchors. Strong GDP growth combined with low unemployment often coincides with a positive output gap, while weak growth and rising unemployment often coincide with a negative gap.
| Year | U.S. real GDP growth | U.S. unemployment rate | Typical output-gap reading |
|---|---|---|---|
| 2021 | 5.8% | 5.3% | Rapid reopening growth often pushes measured output above short-run trend |
| 2022 | 1.9% | 3.6% | Growth slowed, but tight labor markets can still imply little slack |
| 2023 | 2.5% | 3.6% | Moderate growth with low unemployment can support a near-zero to mildly positive gap |
Those growth figures come from BEA annual real GDP changes, while the unemployment rates are consistent with BLS annual averages. You should treat them as context rather than a one-for-one mapping. A country can post healthy growth while still having a negative output gap if trend growth was stronger, and it can show a positive output gap during modest growth if capacity has weakened.
Important limitations of the HP filter
Although the HP filter is popular, it is not perfect. One of the biggest issues is end-point bias. The trend estimate near the beginning and end of the sample is less reliable because the algorithm has fewer surrounding observations to anchor the smooth path. This matters in real-time policy work because the current quarter is exactly the point analysts care about most.
- End-point problem: recent observations can shift the estimated trend significantly when new data arrive.
- No structural theory: the filter is statistical, not a full economic model of productive capacity.
- Lambda sensitivity: different lambdas can produce meaningfully different gap estimates.
- Revision sensitivity: GDP revisions can alter both the trend and the cycle.
For high-stakes policy analysis, many institutions complement HP filter estimates with production-function models, labor market indicators, capacity utilization, and inflation dynamics. In other words, the HP filter is often a starting point, not the final word.
Best practices when you use tsa.filters.hp filter to calculate output-gap in Python
- Use seasonally adjusted data whenever possible.
- Prefer real variables over nominal variables for macro gap analysis.
- Log the series if all values are positive and you want percent-style interpretation.
- Choose lambda based on data frequency, then test sensitivity with nearby values.
- Do not over-interpret the last few observations without checking revision risk.
- Compare your gap estimate with labor market, inflation, and capacity indicators.
Authoritative data sources you can use
If you are building a Python workflow for output-gap estimation, these official sources are especially useful:
- U.S. Bureau of Economic Analysis GDP data
- U.S. Bureau of Labor Statistics Current Population Survey
- Congressional Budget Office economy and budget analysis
These sources let you validate your own calculations, compare trend estimates with official assessments, and build a more robust Python pipeline around the HP filter.
When to use the HP filter and when to go beyond it
The HP filter is excellent when you need a fast, transparent, and reproducible decomposition of a macro time series. It is ideal for classroom work, exploratory analysis, dashboarding, and business-cycle visualization. It is also useful when you want a clean way to compare cyclical positions across countries or sectors with similar data structures.
However, if your goal is to estimate true potential output for policy simulation, debt sustainability work, or inflation forecasting, you should usually go beyond the HP filter. Structural models, state-space models, Kalman filtering, and production-function approaches can provide richer estimates, especially when you need a real-time measure that blends multiple signals.
Bottom line
To use tsa.filters.hp filter to calculate output-gap in Python, the practical recipe is simple: collect a clean time series, decide whether to log it, apply the HP filter with an appropriate lambda, and interpret the cyclical component as the gap from trend. The calculator on this page reproduces that logic in an interactive format, helping you test sample values before you move to a full Python notebook or production workflow.
If you are just getting started, begin with quarterly real GDP, use a lambda of 1600, and inspect both the level chart and the gap chart together. That combination gives you a useful first-pass estimate of economic slack while keeping the workflow transparent and easy to explain.