מסמך טכני מיומן המחקר · באנגלית, כפי שנכתב בשלב העבודה
Step 290: Israeli investment estimates within a single publication vintage
Step 289 tested two ways of linking vintages. This check estimates the same model using the older release on its own and removes the pandemic quarters as an endpoint check. The recent release is the v288 baseline. All slopes are percentage points of nominal GDP per unemployment percentage point. Intervals are approximate 95% Newey–West intervals and the model controls for contemporaneous and lagged real GDP growth.
| Nonhousing investment sample | One-quarter slope [interval] | Four-quarter slope [interval] |
|---|---|---|
| CBS 1995–2020 release, observations 2012Q1–2020Q3 | 0.53 [−0.58, 1.64], n=33 | −1.13 [−2.87, 0.62], n=27 |
| Same release, through 2019Q4 | 0.59 [−0.43, 1.60], n=30 | −0.66 [−2.38, 1.05], n=24 |
| CBS 1995–2023 release, observations 2014Q1–2023Q4 | 0.22 [−0.48, 0.92], n=38 | −0.29 [−0.89, 0.30], n=32 |
The opposite point-estimate signs across horizons occur in both single-vintage samples. All intervals include zero. The endpoint comparison is descriptive and cannot isolate a pandemic effect. The older full sample also weakens the large negative estimate in the shorter overlap-only v289 diagnostic. These results do not support selecting an Israeli calibration.
Historical source status
- The CBS 1995–2023 publication says the web tables extend to 1995. The printed quarterly tables 35 and 38 used here start in 2014. Direct attempts to retrieve candidate table downloads did not yield verified workbooks.
- The CBS 1995–2011 release contains earlier quarterly tables. The searchable extraction of table 33 omits some quarter labels and repeats numeric rows, for example the 2008 Q2 and Q4 entries. This extraction was excluded from estimation. Table 33 also reports chained real values, whereas the target outcome uses nominal shares.
- The acquisition target is one release of the original quarterly estimates, preferably the tables 35 and 38 spreadsheets. Required columns are current-price GDP, chained real GDP, current-price total fixed capital formation, and current-price residential construction. Before estimation, verify unique consecutive quarters, compare nominal fixed investment across both tables, reconcile annual nominal totals against the same release, and document any labor-force series breaks.
Reproduce the added checks with python scripts/israel_investment_vintage_v289.py under the keys old_full and old_through_2019.