AI review — page 14

claude-opus-4-8 · prompt v7 · 2026-08-21T10:45:04+00:00 · applied: yes · changed: yes

Screenshot sent to the model (reading-order tags burned on)

Instructions (system prompt)

The screenshot has annotations burned onto it that are NOT part of the document:
- a small red numbered tag at the top-left corner of each item, showing that item's position in the OCR-determined reading order (the same order the items appear in the json data);
- a red-and-white dotted outline around each item, showing the area the OCR detected for that content block.
Use the tags and outlines to see the detected reading order and item boundaries directly on the page, and judge that sequence against how a human would naturally read it. Ignore the annotations when checking text fidelity — they overlay the content, they are not content.

For the given screenshot of a PDF's page and the attached json data, I want you to perform the following tasks in order...

Review the reading order set out in the json data and compare to what a natural reading order for that of a human would be by looking at the screenshot. Decide on any changes and re-arrange the items for the most logical reading order.
Look at all text for each item and correct any extraction errors like missing words, spelling mistakes etc.
Look at each item in the JSON and ensure that the OCR process has identified the item as the correct type: text, list item etc.
Look for any text that duplicates: if two items next to each other contain the exact same text but the text only appears once on the screenshot, one of them is an extraction error — keep the item whose box matches where the text is visible and flag the other for removal.
Make amendments as you proceed through the items and list of instructions.
Reading Order Advice: If there is columns with headings and text, I wouldnt expect the reading order to jump from one heading straight to another if there is text associated with that heading.

Return format: Give me the json data back only, with the amendments you make.

Mechanical notes (so your answer can be applied automatically):
- Each item has an `id` — keep every item and its `id` exactly as given; never invent, drop or duplicate ids. Re-arranging means changing the position of items (and their nesting) in the arrays.
- Never move text (or a type) from one item to another: each item's coordinates travel with its id, so to change reading order you must move the whole item object, and text amendments must be in-place corrections of that item's own text.
- To flag a duplicate item, keep it in the array and add `"remove": "duplicate"` to it — never just delete it (deleted items are restored automatically).
- ids are opaque labels, not sequence numbers: never renumber them. After a removal or re-arrangement, every remaining item keeps the exact id it came with, even if the ids no longer look sequential.
- Text may contain `[pN.M…]` placeholders marking where an inline formula belongs — treat them as part of the text and leave them exactly where they are.
- `box` is [left, top, right, bottom] as percentages of the page from the top-left corner; return it unchanged.
- Respond with raw JSON only: no code fences, no commentary, same shape as the input (`{"items": [...]}`).

User message (json data sent)

Page 14 json data:

{"items":[{"id":"p14.1","type":"text","box":[11,9,88,14],"text":"with search frictions as harmonised as possible with the standard model, which does not distinguish the labour market for different groups of households. The parameter governing the production of labour services is in line with Bodart et al. (2006).","children":[{"id":"p14.1.1","type":"link","box":[62,12,73,14],"text":"Bodart et al."},{"id":"p14.1.2","type":"link","box":[73,12,78,14],"text":"(2006)."}]},{"id":"p14.2","type":"text","box":[11,14,89,26],"text":"Matching efficiency and vacancy posting costs are obtained by targeting matching probabilities of workers and firms. Matching probabilities of workers come from the OECD and Eurostat data for unemployment duration by educational attainment for 2017 and 2018 using the method of Shimer (2012).[p14.2.9] The matching probability for firms is based on the estimates for the US in Den Haan et al. (2000) and Stähler and Thomas (2012) for Europe. These two probabilities determine matching efficiency and (indirectly) vacancy posting costs (the latter being lower for HtM households).","children":[{"id":"p14.2.1","type":"link","box":[46,19,52,21],"text":"Shimer"},{"id":"p14.2.2","type":"link","box":[53,19,58,21],"text":"(2012)."},{"id":"p14.2.3","type":"link","box":[59,19,61,21],"text":"10"},{"id":"p14.2.4","type":"link","box":[53,21,67,22],"text":"Den Haan et al."},{"id":"p14.2.5","type":"link","box":[68,21,73,22],"text":"(2000)"},{"id":"p14.2.6","type":"link","box":[78,21,88,22],"text":"Stähler and"},{"id":"p14.2.7","type":"link","box":[12,22,19,24],"text":"Thomas"},{"id":"p14.2.8","type":"link","box":[20,22,25,24],"text":"(2012)"},{"id":"p14.2.9","type":"formula","box":[11,14,89,26],"text":"¹⁰"}]},{"id":"p14.3","type":"text","box":[11,26,89,38],"text":"Separation rates are obtained by matching unemployment rates, which are OECD data averages from 2004-2019. Consistently with job finding probabilities, we assume that the unemployment rate for non-Ricardian households is the unemployment rate for persons with educational attainment below upper-secondary. Matching the unemployment rates results in break-up rates for HtM households exceeding those of Ricardian households. Bargaining power has been set to 0.5, for both groups of households, in line with the literature."},{"id":"p14.4","type":"text","box":[11,38,88,50],"text":"To harmonise the models, we match the wage and total labour services from the search model with those in the standard model. To match the wage, we use the replacement ratio in the search model (this indirectly influences wage through the workers' outside option). This results in replacement ratios that are close to the OECD estimates for Ricardian and HtM households, which are typically around 0.5 and slightly higher for the HtM households.[p14.4.2] To match labour services, we use the total time endowment of households.","children":[{"id":"p14.4.1","type":"link","box":[41,46,43,48],"text":"11"},{"id":"p14.4.2","type":"formula","box":[11,38,88,50],"text":"¹¹"}]},{"id":"p14.5","type":"text","box":[11,51,88,56],"text":"[p14.5.5]As described in Gomes et al. (2023) and Herman and Lozej (2023), we assume that non-Ricardian households have below upper-secondary education and Ricardians have above upper secondary education. The estimates are population-weighted across countries.","children":[{"id":"p14.5.1","type":"link","box":[28,50,39,53],"text":"Gomes et al."},{"id":"p14.5.2","type":"link","box":[40,50,44,53],"text":"(2023)"},{"id":"p14.5.3","type":"link","box":[49,50,64,53],"text":"Herman and Lozej"},{"id":"p14.5.4","type":"link","box":[65,50,69,53],"text":"(2023),"},{"id":"p14.5.5","type":"formula","box":[11,51,88,56],"text":"¹⁰"}]},{"id":"p14.6","type":"text","box":[11,56,88,61],"text":"[p14.6.1]Concretely, replacement rates for one-year horizon of unemployment duration for Ricardian households and two-year horizon of unemployment duration for non-Ricardian households, given that they are more likely to be long-term unemployed.","children":[{"id":"p14.6.1","type":"formula","box":[11,56,88,61],"text":"¹¹"}]},{"id":"p14.7","type":"footer","box":[49,94,51,95],"text":"13"}]}